of Docum entation A Context-based Study of Serendipity in Information Research among Chinese Scholars

Purpose 
 
 
 
 
The current understanding of serendipity is based primarily on studies employing westerners as the participants, and it remains uncertain whether or not this understanding would be pervasive under different cultures, such as in China. In addition, there is not a sufficient systematic investigation of context during the occurrence of serendipity in current studies. The purpose of this paper is to examine the above issues by conducting a follow-up empirical study with a group of Chinese scholars. 
 
 
 
 
Design/methodology/approach 
 
 
 
 
The social media application “WeChat” was employed as a research tool. A diary-based study was conducted and 16 participants were required to send to the researchers any cases of serendipity they encountered during a period of two weeks, and this was followed by a post-interview. 
 
 
 
 
Findings 
 
 
 
 
Chinese scholars experienced serendipity in line with the three main processes of: encountering unexpectedness, connection-making and recognising the value. An updated context-based serendipity model was constructed, where the role of context during each episode of experiencing serendipity was identified, including the external context (e.g. time, location and status), the social context and the internal context (e.g. precipitating conditions, sagacity/perceptiveness and emotion). 
 
 
 
 
Originality/value 
 
 
 
 
The updated context model provides a further understanding of the role played by context during the different processes of serendipity. The framework for experiencing serendipity has been expanded, and this may be used to classify the categories of serendipity.

Research into serendipity has been ongoing since the term "serendipity" was first coined by Horace Walpole in 1754, in reference to the Three Princes of Serendip, who were always making discoveries by accident. Studies relating to serendipity can be found in various disciplines, including information studies (Foster and Ford, 2003),  (Sun et al., 2011), social science (Merton, 2004), career research (Bright et al., 2005), arts and humanities (Delgadillo and Lynch, 1999), psychology (Heinström, 2006), organisation (Cunha et al., 2010), and medicine (Allegaert, 2013).
However, regardless of the increasing interest in the understanding of serendipity, an interesting discovery revealed from our review of current studies is that the proposed frameworks or theoretical models for serendipity were formed primarily on the basis of taking Westerners as the research subjects (e.g. Makri and Blandford, 2012a;McCay-Peet and Toms, 2015). This led to our thinking about whether or not these research findings would be pervasive under different cultures, such as in China.
In addition, although a number of theoretical models for serendipity have been put forward by various researchers (Erdelez, 2004;Makri and Blandford, 2012a;McCay-Peet and Toms, 2015;Rubin et al., 2011), very few of these studies have examined the occurrence of serendipity from a systematic perspective of context. The empirical studies by Points et al. (2015) showed contextual factors such as location, activity and focus can influence a user's experience of serendipity. Kefalidou and Sharples's (2016) study also found that time, location, and the content of a text message can also impact a user's experience of serendipity. Serendipity, as part of a wider behaviour model, is considered as "the product of context" (Foster and Ellis, 2014, p.18), and the role of context in fostering serendipity deserves to be understood in its own right.
We previously undertook a mobile-diary study to understand serendipity among a group of British researchers (Sun et al., 2011). Eleven British scholars participated in that study, and we received 23 serendipity cases within one week. The outcome of the study was that we identified the perception of serendipity among these British researchers and made an initial probe into the role of context in serendipitous experiences (See Figure 1). We identified two different levels of abstraction that can lead to a positive outcome during a serendipitous experience, including level of abstraction 1 "the unexpected finding of information" and level of abstraction 2 "making unexpected connections between pieces of information", and in some cases level of abstraction 1 can lead to level of abstraction 2. Context played a role in affecting the experiencing of serendipity, and it has been examined from the three perspectives of: people (active or less active), temporal factors, and environment (i.e. working environments, places, and changing environments). We then adopted Schmidt's (2000) context model to denote the relationship between the role of the having an influence on participants' readiness to experience serendipity, and thereby failed to make a further examination of how these elements would act during the separated processes of serendipitous encountering.
Based on our previous study and the inspiring achievements made in the field of serendipity study in recent years, we carried out a follow-up empirical study among a group of Chinese scholars with the following research aims: • To identify whether the current understanding of serendipity can also be adapted to Chinese scholars; • To further investigate the role played by context during the different processes of experiencing serendipity.

Context in serendipity research
In recent decades, a number of researchers have performed different studies with respect to serendipity, although these researchers have not reached a consensus on the definition of serendipity. For example, van Andel (1994) defines serendipity as "the art of making an unsought finding", while Fine and Deegan (1996) give the definition of serendipity as "the unique and contingent mix of insight coupled with chance".
More recently, McCay-Peet and Toms (2015) contend that serendipity is "an unexpected experience prompted by an individual's valuable interaction with ideas, information, objects, or phenomena", while the term serendipity is defined in Björneborn's (2017) paper as "what happens when we, in unplanned ways, encounter resources (information, things, people, etc.) that we find interesting". Rather than giving a definition, Makri and Blanford (2012) identified three key elements for serendipitous encountering: unexpectedness, insight and value.
However, regardless of the various definitions, it is well accepted by information researchers that serendipity is an integral part of information behaviour, and "context" is a significant concept when studying information behaviour, as argued by Case and The seeker-whether actively looking for information or receiving information through serendipity-exists in an environment that partially determines, constrains, and supports the types of needs and inquiries that arise. The seeker also has his or her own memories, predispositions, and motivations-an internal environment of influence. (Chapter 3, p. 48) and Context determines much of a person's perceptions throughout the [information seeking] process, and it affects one's choice of sources and meanings. (Chapter 11,p. 351) Björneborn's (2017), who adopted the theory of affordances from Gibson (1977), also argues that serendipity can be viewed as an affordance, which should never reside inside the environment alone nor inside people alone, but should be viewed as the relational phenomenon between people and a given environment. In a similar vein, after a review of serendipity studies in information research, Foster and Ellis (2014) concluded that serendipity does not exist within a vacuum, but is "the product of context" (p.18). Some empirical studies also demonstrate contextual factors affecting an individual's experiencing of serendipity. For example, through a "Wizard of Oz" approach, where users received text messages/suggestions from a group of "wizards" based on users' notes in an app "SerenA", Points et al. (2015) found that those contextual factors such as location, activity and focus can influence a user's experience of serendipity. Similar findings can be found in another paper (Kefalidou and Sharples, 2016), where the contextual factors such as time, location, and the content of the text message can impact a user's experience of serendipity. McCay-Peet and Toms (2015) have found that those environmental factors which are triggerrich, enabling connections and leading to the unexpected can help users facilitate serendipity in a digital environment. Such ongoing research findings provide substantial evidence that context does play a vital role in people's experience of serendipity.

A Further Discussion of Context
From a review of the existing studies on the issues of context and serendipity, it is evident that none of them have systematically discussed the term "context", nor how it may influence the different processes during a serendipitous encounter. Björneborn  Foster and Ellis (2014) argued that "even context is debatable and has been the subject of exploration in its own right" (p. 18). Case and Given (2016) considered context to be "ill defined", but also highlighted its important role when integrated in human information behaviours. Taken together, we believe there is a need to probe this special term "context".
Based on a review by Courtright (2007), the study of "context" in information science has shifted from a "system-centred" to a "user-centred" stance. A previous "system-centred" view regards context as an "objective reality" (Talja, 1997), which has served as a backdrop for those environmental factors or variables that exist objectively around the information actor, and can therefore be enumerated by the researcher. Such a view of context is also labelled as "objectivist" (Talja et al., 1999), which presents context as a set of entities that can be conceptualised independently to influence a participant's information practices (e.g. temporal or spatial conditions, problem situations, etc.). However, taking only those environmental variables into the consideration of context fails to shed light on the variability among actors in the same or similar settings. The information actors can carry out actions independently and differently in response to the variability of the environmental factors in their information practices. Therefore, an increasing number of researchers have now attempted to examine the role of context from the viewpoint of the information actor.
This "user-centred" view emphasises the role of information actors during their information practices, and considers the information activities in relation to the contextual variables and influences. Various models have been constructed to support this kind of view, such as Wilson's (1981) information seeking model where an individual's physiological, affective and cognitive needs are located in the concentric layers at the root of motivation towards the information seeking behaviour. Foster's (2004) nonlinear model for interdisciplinary information-seeking also highlights information seekers' feelings and thoughts, coherence, knowledge and understanding as the internal context to influence information seeking behaviour. Although such a person-in-context stance is being accepted by more and more researchers, there are also critics who argue that these models do not account adequately for the mutual interactions of contextual factors, especially the social interactions. Each individual is conceptualised as a social actor (Lamb et al., 2003) and knowledge as inherently social (Talja, 1997). Therefore, information actors should construct information not interactions. Rather than simply observing the information actors' behaviours or recording their views, the relevant discourse should also be taken into consideration when trying to gain an understanding of the role of context in information research (Given, 2002;Sundin, 2002;Talja et al., 2005). Taken together, Courtright (2007) suggests the combination of multiple methods to paint a comprehensive portrait of context, which should not only try to capture any environmental variables around the information actor, but should also try to understand their mind-sets and follow the links across their multiple social settings.
Following the above discussion, in this paper we discuss the environmental variables as the "external context", the mind-sets relating to the role played by the information actor as the "internal context", and the social settings around the actor as the "social context".

Existing Serendipity Models
Although the study of serendipity is still an emerging research discipline, several studies in information research have explored how serendipity happens, and theoretical models have been designed by the researchers. We reviewed the six existing models for serendipity, five of which are process-oriented while the sixth is based on the essence of serendipity.

Process-orientated models
The first model designed to help with an understanding of serendipity was proposed by Erdelez (2004), who also labelled serendipity as "information encountering", a specific type of opportunistic acquisition of information. The study was undertaken in a controlled environment where users were asked to actively look for information relating to a particular foreground problem, but where they actually encountered information relating to a background problem. According to this model, the information encountering process is divided into five stages: noticing, stopping, examining, capturing and returning. A user's current searching behaviour with regard to the foreground problem is interrupted when he notices the information related to the background problem. The user then stops to examine this information, captures any useful details and finally returns to the search relating to the foreground problem. McCay-Peet and Toms (2010) adapted Cunha's (2005) conceptual model of the serendipity process in organisational management, and identified the process of serendipity as follows: while searching for a solution to task A, with certain precipitating conditions a person perceives a trigger and then sparks a bisociation between disparate, previously unconnected pieces of information and finally this leads to an unexpected solution to task A, or even to a new task B. The most salient point of this model is the precipitating condition (Cunha, 2005) which shows that, to some extent, serendipity can be guided with appropriate strategies. This model has been updated recently by combining several other models with respect to serendipity (McCay-Peet and Toms, 2015). The process of a serendipitous experience is redefined as a combination of seven elements, namely, trigger, delay, connection, follow-up, valuable outcome, unexpected thread and the final perception of serendipity. In addition, they argued that, unlike other elements, the elements of delay and follow up "do not have to happen for perception of serendipity to occur".
The remaining two models are more focused on the mental processes of individuals who have had serendipitous experiences. Lawley and Tompkins (2008) considered serendipity as "the whole shebang" with six components including the prepared mind, an unexpected event, recognised potential, seizing the moment, amplifying the effects and evaluating the effects. They argued that following the removal of any of the six components and the iterative circularity from recognising potential to amplify the effects, the process would no longer be regarded as serendipity. Makri and Blandford (2012) developed their serendipity model based on semi-structured interviews with 28 interdisciplinary researchers. Their findings suggested that unexpected circumstances and insight could stimulate a person to make new connections with an iterative process by projecting the potential value of an outcome and further exploring the value to gain a valuable, unanticipated outcome.
Apart from the five process-oriented models, there is another important model which depicts the essence of serendipity. Rubin et al. (2011) employed a selective blog minding method by analysing 56 blog entry accounts of chance encounters, from which they identified four key facets which can be used to facilitate serendipity: a prepared mind (including a prior concern and previous experience), an act of noticing (the ability to notice the provided clue), chance (an accidental or unplanned encounter with the find) and a fortuitous outcome (unexpected benefits linked to the find). An individual may conclude whether or not an event should be regarded as serendipity by a reframing of these four facets.
On further reflection of these models, we have found that from a "user-centred" point of view, these models all partly refer to the contextual factors, especially those relating to the information actor, as is listed in Table 1. Erdelez's model requires users to notice the background problem, and such an "ability to notice" can be considered as part of the internal contextual factors that affect an individual's experiencing of serendipity. In addition, her model is useful for understanding part of the process postencounter, but it fails to cover what happens beforehand (e.g. whether or not external factors played a role to trigger the encounter). McCay-Peet and Toms's (2010) model identified "precipitating conditions" as "active learning" (internal context) and "social networks" (social context), and as a requirement for a "trigger" (e.g. text, images, audio) to facilitate serendipity. However, as an early model in knowledge work, this model fails to look into how the "precipitating conditions" would impact the process of serendipity. In their updated model (McCay-Peet and Toms, 2015), they further highlighted how the "trigger" confirms the "noticing" element of the process of serendipity, and identified the three forms of triggers as "verbal" "textual" and "visual". They further proposed different external factors (trigger-richness, highlight triggers, enabling connections and enabling capturing) and internal factors (openness, a prepared mind, the ability to make connections) that may influence the perception of serendipity, but these factors, especially the external factors, are not discussed from the perspective of context. The other two mental-process models focus mainly on the perceptual process required for a serendipitous episode, and also discussed some contextual factors. For example, Lawley and Tompkins (2008) considered a "prepared mind" (internal context) and an "unexpected event" (external context) as necessary components in a serendipitous episode, while similarly, Makri and Blandford (2012a) considered how "unexpected circumstances" (external context) and "insight" (internal context) can lead to making new connections, and they also found that, although not directly reflected in their model, their participants' moods or feelings (internal context) can impact the openness to making connections. Similarly, chance (external context), a prepared mind, an act of noticing and surprise (internal context) can all be considered as contextual factors that are referred in Rubin et al.'s (2011) However, a systematic discussion from the perspective of these contextual factors cannot be drawn from these existing studies.

Research Method
We developed a mobile-diary method for our study in the UK to help participants capture their serendipitous experiences in a comparatively natural setting. Satisfactory feedback was received via the mobile diary application, with 23 serendipitous cases reported by 11 British participants within a one-week period. One major concern raised by some of the participants in that study was that they were unwilling to carry around an additional mobile device, and this undoubtedly affected their serendipitous experiences. To overcome this particular drawback, in this study we used the social media platform "Wechat" to replace the diary application. The main reasons for choosing "Wechat" were: 1) it covers similar functions to our diary application, and different types of data can be recorded and transferred (i.e. text, video, audio, and image); 2) participants had no concerns about portability problems, and no additional package needed to be installed on participants' own mobile phones, as they were all frequent users of "Wechat", and were quite familiar with its functions; 3) "Wechat" is a social media platform, so it also has the function of allowing direct communication between participants and the researchers, and if participants had any problems during the experiment period, they were able to send messages to the researchers and receive immediate responses; 4) it had the advantage of allowing the researchers to send a "reminder" to participants each day, to help to make them aware they were in an experiment situation.

Participants
16 Chinese PhD students (eight males and eight females) were recruited to take part in this follow-up study, with each participant having had at least 12 months' research experience. We chose PhD students mainly because: 1) following the research findings of Foster and Ford (2003) which showed that serendipity is experienced widely among researchers, and PhD students are a group of scholars dedicated to research projects who are easy to access; 2) our previous study recruited 2) Pre-interview. Each participant was invited to a short interview (around 30 minutes) before the empirical study. The research purpose was introduced, and participants were invited to collect any cases they considered as serendipity during a period of two weeks, either on the Web or as part of their daily activities (e.g. reading, research, and socialisation). In addition to introducing the research purpose to each participant, we also conducted two additional operations during the pre-interview, as follows: (a) Each participant's initial understandings of serendipity were collected. During the interview process we found that each participant reported that this was the first time s/he had heard about the concept of serendipity. To better support the study, we then carefully introduced participants to this concept. First, we presented the definition of serendipity from the Oxford Concise English Dictionary: "the occurrence and development of events by chance in a happy or beneficial way". We then provided participants with the following example from the pilot study: We took care to highlight to participants that this example was simply provided to help them to understand the concept of serendipity, and by no means to restrict them to a particular type, stressing that there are various examples. They were instructed that if they experienced any encounters which they considered as serendipity, they should send the researchers a relevant message.
(b) Participants were introduced to the group created using the social media tool "Wechat" and its functions. To achieve better research results, we designed a specific interface and instructed our participants on how to use its functions ( Figure 2). A detailed description of the interface is provided in Appendix A.
{Insert Figure 2 here} 3) Two-week study. A lesson we learnt from our previous study is the necessity to extend the experimental time window to give our participants sufficient time for potential encounters with serendipitous experiences (one participant failed to send us any information and argued that the time available was insufficient for him to encounter serendipity). Thus, we set the experiment time period at two weeks for this study. Participants were required to use the tool provided to record their serendipitous experiences, and return them to the researchers within two weeks. All the sent data was only visible to the researchers. In addition, at approximately 10:30pm each day, a reminder message was sent to each participant by the researchers to better provide them with a research context (Figure 2-c). The time chosen for sending the reminder was based on the pilot study and observation of the routines of most participants.

4) Post-interview.
Each participant was invited to a post-interview at the end of the study. The interview was conducted within one week and lasted for approximately one hour with each participant. It was semi-structured and qualitative in nature and centred on participants' recorded serendipitous encounters, as well as participants' experiences of the research method.

Data Collection
Two types of data were collected: the recorded diary data of the participants' serendipitous experiences and the post-interview data. We received a total of 62

Data Analysis
The data we collected are qualitative in nature. A Thematic Analysis (Braun and Clarke, 2006) was conducted to code the interview data. We first followed a topbottom theoretical thematic analysis to investigate how the participants experienced serendipity. We began this part of the coding by identifying the themes drawn from our previous study, where we identified the nature of serendipity according to two different levels of abstraction, and the value of serendipity. The first level identified the "unexpected finding of information" by considering different combinations of three components: whether the information was directly related to the activity being undertaken by the individual (non-activity-based vs. activity-based); whether or not the information encountered was unexpectedly valuable to the encounterers (unexpectedly valuable or not); and whether the information was from an unexpected or likely source. The second level identified the making of unexpected connections between different pieces of information, people and ideas.
We then used a bottom-top inductive thematic analysis to identify any contextual factors which existed in the serendipity cases. Initially, we identified a number of categories, including the time for experiencing serendipity (i.e. a.m., p.m., and across time periods), the different locations in which serendipity occurred (e.g. office, dormitory, classroom, , library, etc.), the different activities during which serendipity was experienced (e.g. travelling, surfing the Internet, attending seminars, talking to classmates, talking to friends, etc.), and a category more related to an individual's cognitive or psychological characteristics, such as memories, an amount of thinking, expertise, previous needs, instantly raised needs, and emotions (see Figure 3, for examples of the coding for the pilot study case). We then compared the categories and grouped those categorised with overlapping meanings into possible themes. As a result, we concluded this layer of analysis with the three major themes of: external context (i.e. time, location, and personal status), social context (e.g. different social counterparts) and internal context (i.e. precipitating conditions, sagacity/perceptiveness and emotions). It should be noted that the precipitating  After we finished coding the themes of the contexts, especially the internal context themes of the precipitating conditions, we carried out a full review and found that our original coding of the first level "unexpected finding of information" could also be considered as a process of making connections between the encountering and the precipitating conditions. As a result, we re-coded this part of the framework into the three different themes of unexpectedness, connection-making and value, which were further expanded into the sub-themes of "unforeseen means of encountering information", "unexpected content of the encountered information" and "both". The theme of "connection-making" was further expanded into "connection-making between unexpectedness and visceral needs", "connection-making between unexpectedness and conscious needs" and "connection-making between unexpectedness and previous experience/knowledge", while the theme of "value" was expanded into the sub-themes of "substantial value" and "emotional value". This will be discussed in detail in the following sections.

Participants' Perceptions of Serendipity
By analysing the 62 reported serendipitous cases, we found that the Chinese scholars conformed to the framework of experiencing serendipity according to the three main processes of: encountering unexpectedness, connection-making and finally leading to a valuable outcome.

Unexpectedness
Three different channels were identified from our empirical study to facilitate the likelihood of encountering unexpectedness during the new study: (i) Any unforeseen means by which a participant encounters a piece of information. An example, which can better explain our identification of this element, is provided in the following case: The participant (P3) reported she had learnt about the principles of fire extinguishers from a talk during a training session. She then raised the need to collect relevant information about fire extinguishers around her lab setting (e.g. where they are located in the lab, and how to use them). However, this need was not addressed at that time and the participant forgot to address the need after the talk had finished. As explained by the participant during the interview, "I just thought about it in my mind and didn't write it down during the talk. Actually, when the talk finished, I just forgot about it". It was not until she accidently attended a related "fire alarm introduction" that she realized the need again, and found the answer to this need during the presentation. We argue that the answer to the participant's need was not unexpected to her, but the way that she received the answer made her feel it was "unexpected", as attending such a presentation was not her original schedule, "I even didn't know about such a presentation, but one of my friends just asked me to accompany her".
(ii) The content of the encountered information brings unexpectedness. In some other cases, it is the content of the information that leads to a sense of unexpectedness: From this response, it is evident that both the useful information she received from the seminar, and the way she received the information (by attending the seminar accidentally) functioned in her coming across this serendipitous experience.

Connection-making
We have previously found that connections can be made between different pieces of information, people and ideas (Sun et al., 2011), and it is a level of abstraction that can lead to a positive impact. In this paper, we have further expanded this process of connection-making by identifying the different internal contextual factors of precipitating conditions: • Connections made between unexpectedness and visceral needs.
• Connections made between unexpectedness and conscious needs.
• Connections made between unexpectedness and previous experience/knowledge.
The term "visceral need" and "conscious need" originates from Taylor's (2015) work, which characterized four different levels of information needs during the interaction between an information seeker and a librarian. An information seeker may begin with an unexpressed need in mind (what Taylor calls a "visceral need"), and then such a need becomes "conscious" with accumulated information (e.g. by talking to the librarian), and turns to a "formalised need" with a qualified and rational statement, which finally leads to a "comprised need" that can be presented to information systems. Although Taylor's framework is usually used to describe the negotiation process between an information enquirer and an information specialist, which is not exactly the process of encountering a serendipitous episode, we do find that the characteristics of such "visceral needs" and "conscious needs" conform to the situations reflected in the participants of our empirical study. For example, Taylor argues that a "visceral need [not] existing in the remembered experience of the enquirer, [probably] is inexpressible in linguistic terms, [and can] change in form, quality, concreteness, and criteria as information is added." In our study, we also found some participants did not raise a need, which was not previously in their memory or experience, until they encountered unexpected information. While a "conscious need", as described by Taylor, is a "within-brain description", it is quite similar to a previously unaddressed concern/question by a participant. This part will be discussed further in Section 5.3.1.

Value
Our participants pointed out that they would only consider any unexpected experiences as serendipity if they offered them some form of benefit. Two types of value arose from our study: substantial value and emotional value. Substantial value refers to a value that brings beneficial results or outcomes to the participant (e.g. finding the answer to a previous concern), while emotional value refers to a value caused by an emotion which is aroused when a participant encounters unexpected information (e.g. the emotional satisfaction of recalling previous memories). This part will be discussed further in Section 5.3.3.

The Role of Context
Context plays a significant role in nurturing serendipitous experiences, although it has often been neglected in previous studies. As a complement to our previous work (Sun et al., 2011), in this paper we have identified the conception of context more comprehensively, based on the discussion in the background section, including external context, social context and internal context, and different contextual factors were identified from the empirical study.  (Fröberg, 1977), and even the cognitive and evaluative efficiency of individuals (Natale et al., 2003). Our participants also reported that they were more engaged in different activities in the afternoon, as a result of which it was also more likely that they would encounter serendipity, as explained by one of the participants: Personally speaking, I find myself more conscious about the concept of serendipity in the afternoon or evening than in the morning, and normally I'm more engaged in the afternoon. So I think that's the reason why I always send you messages in the afternoon. (P 4) (2) Location. Some locations (e.g. libraries) are richer in resources (e.g. books) than other locations (e.g. canteens). Therefore, it is intuitively sound to assume that locations may influence the occurrence of serendipity. 58 of the cases reported by our participants indicated where their serendipitous experiences had taken place (the location of the remaining four cases could not be recalled). Our data showed that 29 cases happened in an office environment, followed by 11 cases in a seminar room, six cases in a dormitory and other random places (laboratory, café, library, etc.).
Following a further look into the office environment, which produced most serendipity cases during our study, we found that there were three possible reasons that contributed to the encounter of serendipity: The office environment is resource-rich, including posters, notifications, different online libraries, etc. Where more information is presented to a participant, s/he will have an increased possibility of experiencing serendipity; • It provides an interdisciplinary social setting. The work settings for the participants was interdisciplinary, so a participant from mechanical engineering would be sitting in the same office as colleagues from other backgrounds, such as chemical engineering, architecture, or computer science, etc. Foster and Ford (2003) provided several examples of experiencing serendipity among interdisciplinary researchers, and similar cases were also collected in our study.
For example, one participant (P 2) from mechanical engineering accidently learnt about a new image-searching engine "TinEye" during a break, when talking with a colleague from computer science, which helped him locate the resource literature for an image he had used in his writing. Another participant (P 6, design background) also encountered useful information about using Nvivo to help his data analysis, of which he was not previously aware, from a casual conversation with a colleague with an HCI background; • It makes it easy to get access to different resources. Consider the following example. One participant (P 7) sent us a case in which he happened to encounter a method of "histogram equalization" from a blog when he was browsing the Internet in the office. He then conducted a further search into this method (by referring to Wikipedia and other relevant literature), which enabled him to understand this method and recognise its value (i.e. it could be used in his own research). Compared to other environments, such as a laboratory or a café, it is obvious that the accessibility of resources (e.g. licenses to libraries) impact a participant's judgement of the value of the encountered information.
(3) Status. Status here refers primarily to a participant's commitment to certain ongoing activities. We have identified three different types of personal status, as follows: • • Seminar: the participant was attending a seminar, a lecture or a presentation where the participant was a student or a listener.
• Working/Studying: the participant was in an intense and focused state carrying out research-related tasks.
According to Table 3, among the three different types of personal status, participants tended to experience serendipity more often during their "leisure" time.
Studies show that openness and a relaxed setting can facilitate encountering serendipity (McCay-Peet and Toms, 2015;Sun, et al., 2011). Compared to the status of attending a seminar or working/studying, we consider that the participants in the leisure status were more open and relaxed.

Social context
Socialisation has always been considered as a significant factor when it comes to discussions on the role of context (Foster, 2004). In our collected cases, aside from the 37 of 62 cases where the participants had a different personal status, the remaining 25 all took place when they were socialising with others (see Table 4).
{Insert Table 4 here} Table 4 illustrates that participants experienced serendipity frequently when they were socialising with different people, ranging from the familiar (e.g. classmates, colleagues) to the unfamiliar. However, an obvious trend which can be identified from the table is that our participants tended to experience serendipity more often during periods of socialisation with their peers (classmates, colleagues and friends), while only one case occurred during contact with a superior. This differs from our previous study in the UK, where four of the collected serendipity cases under a social context came from communicating with superiors. A possible reason for such a phenomenon is "power-distance", which is a widely-understood cultural difference between the West and the East (Hofstede, 1980), while China has been confirmed as one country with a high power-distance, where students are known to keep a larger interaction distance from their professors (Richardson and Smith, 2007). Therefore, compared to communicating with their superiors, the Chinese participants were more likely to communicate with their peers, leading to more serendipitous encounters.

Internal context
Based on our collected data, we have divided the internal contexts into the following three aspects: precipitating conditions, sagacity/perceptiveness and emotions.

Precipitating conditions
Precipitating conditions refers to the prior conditions for experiencing serendipity.
We have defined such precipitating conditions from a more subjective perspective, which is in relation to an individual's mind-set, covering visceral needs, conscious needs and previous experience/ knowledge. In this case, the participant was initially unconscious of her need to contact an old classmate. It was not until encountering the unexpected information (i.e. her friend was working with him) that she became aware of such a need (to make contact and perhaps collaborate in the future).
-Conscious needs. This indicates that a participant had encountered a need/concern at a previous time (e.g. the need to download a useful paper), but for some reason the need/concern failed to be addressed immediately (e.g. unable to access the data resource). Such a need/concern was resolved when the participant encountered the information unexpectedly, as in the following example: However, unexpectedly, this need/concern was addressed as a result of serendipity.
-Previous experience/knowledge. Several participants also reported that serendipitous experiences were triggered by their previous experience or knowledge: These cases demonstrate that the prior knowledge, interests and personal experiences can be recalled in unexpected ways, thereby contributing to serendipitous experiences.

Sagacity/Perceptiveness
Unexpectedness and precipitating conditions are two necessary components for making connections. However, sagacity or perceptiveness is also required when  (2006) and Rubin et al. (2011) have both considered the act of noticing as an ability to "shift the attention from a primary activity to a clue in the environment", and it is "one of the central elements in incidental information acquisition". Sagacity/perceptiveness is also reflected in our study, as in the following example: The participant experienced different information on three different occasions, and during the interview he emphasised how it was the concatenation of the three experiences that made him consider the whole matter as one piece of serendipity. The sagacity/perceptiveness in this case helped the participant to make connections between a conscious need (strategies to search for academic papers) and the unexpected encountering of information on three different occasions. Apart from this example, during the two-week study, the number of serendipitous cases we collected from each participant ranged from one to thirteen. Such differences in sagacity/perceptiveness are consistent with the discovery by Erdelez (1997) that serendipitous encounterers can range from super-encounterers to occasional encounterers.

Emotions
When a connection is made between unexpectedness and the precipitating conditions, participants switched their attention from the current task to an evaluation of the serendipity they had encountered. Once the value of an encountered experience is acknowledged, serendipity occurs. The term "valuable outcome", which has been highlighted in previous studies (Makri and Blandford, 2012; McCay-Peet and Toms, Rubin et al., 2011), is also manifested in our study as follows: (1) Finding answers to a prior problem or concern. This is most relevant to those serendipities with conscious needs, where the participants are aware of their needs but the answers to prior problems were obtained in an unexpected way.
(2) Providing a potential solution to a need or simply discovering other needs.
This is most relevant to visceral needs, as in the example reported in Case 44, where the unexpected information from the participant's classmate may be useful in the future, but whether or not a desirable outcome can be reached in the future is still unknown.
We identified such a value as "substantial value", as it is most relevant to those need-oriented serendipity cases (either for conscious needs or visceral needs), and it is of substantial benefit for helping participants to address their needs or concerns.
However, apart from "substantial value", we also identified from the study "emotional value" -which highlights the role of emotion-, especially in those cases categorised as previous experience/knowledge-oriented. In both Case 20 and Case 60, the unexpected information triggered a huge emotional response from our participants, and it was because such "emotional value" was so compelling that they came to conclude that the encountered experience was an episode of serendipity.
It should be noted that "emotional value" usually accompanies "substantial value" in need-oriented serendipity cases, such that finding solutions to previous conscious needs, or finding potential benefit from visceral needs often accompanies a positive emotion. However, "emotional value" also functions independently in the experience of serendipity, such as in the previous experience/knowledge-oriented cases (e.g. Case Nevertheless, few studies have investigated how the role of emotional value is embedded in the process of encountering and perceiving something as serendipity.

Comment [t4]:
The whole discussion part is rewritten and complemented. We have removed the culture part throughout the paper, and focused on the finding of context. The connection-making process depends mainly on the information encounterer's sagacity/perceptiveness. However, there is also the situation in which connections are provided by external variables, such as from an interaction partner during a period of social contact. One of the participants (P 14) sent us a case concerned with "how to prevent falling asleep while driving" when she was discussing something with her friend, when she unexpectedly received information from her friend that passengers may read books aloud during the journey, which may interest the driver and prevent boredom. The participant considered the received information to be serendipitous because it was both an unexpected idea and she also applied the idea to her own driving, which turned out to be quite useful. In this case, the participant's process of making connections was simplified thanks to her friend's suggestion. Thus, we argue that the external or social context can sometimes facilitate the process of making connections. Furthermore, the interaction time and activity may also prohibit making connections or evaluating the encountered value. One participant (P 6) reported that he had unexpectedly encountered some information relating to his own research while teaching undergraduates in the laboratory. However, he was too busy answering the problems from the students to think further about the encountered information. It was not until he received the daily reminder that he was able to recall the encountered information from earlier in the day, at which point he became aware of the value of the same. In this case, we argue that the participant's status of being busy (e.g. answering questions from students) may have prohibited his immediate recognition of, or attention to, serendipity at the time at which it occurred.
Emotion played an important role in recognising the value of any encounters.
Makri and Blandord's (2012a) study found that a good mood may help participants to exploit the value of an encounter, while a bad mood may impede such exploitation (p. 694, case UD1 and AD4). In our study, we have identified the value of an encounter as a substantial value and an emotional value, and particularly in any previous experience/knowledge-oriented cases (e.g. recalling good memories, as in Case 20), the emotional value would directly lead participants to consider the encounter as an episode of serendipity. Another point which should be noted is the relationship between emotions and sagacity/perceptiveness (the broken arrow line in Figure 4).
Research from the fields of psychology and neuro-science has revealed that emotions can also impact an individual's cognitive processes, such as their memory, decisionmaking, attention or learning (Schupp et al., 2006;Lerner et al., 2015), which may thus further influence how connections are made between any unexpected information and the precipitating conditions.

A comparison with our previous model
There are two main differences between this updated model and our previous model. The first is that we have found that the process of connection-making is actually pervasive in all cases of serendipity. In our previous model, we identified two different levels of abstraction that can lead to serendipity, and connection-making is the second level of abstraction that can sometimes result from the first level, although we also identified some cases of serendipity which directly resulted from the first level of abstraction. However, by identifying the internal context of the precipitating conditions in this new empirical study, we have found that the original first level of abstraction also involves a connection-making process. For example, we previously The second difference between this updated model and the previous model is that it demonstrates the impact of different contexts during each process of serendipity.
Our previous model mainly discussed the role of context in encountering serendipity from three aspects: the role of people (active or less active), the role of temporal factors, and the role of the environment (i.e. the working environment, places, and the changing environment). We directly adopted Schmidt's context model (2000)   2) The role of temporal factors and the role of the environment were further identified by defining the external context (i.e. time, location, personal status) and social context. The external context and social context are the stimuli for unexpected encountering, and they also affect a participant's connection-making (e.g. facilitation) and recognition of the encountered value (e.g. being busy).
The proposed new context model verified serendipity as "the product of context" (Foster, 2014, p. 18), and these external, social and internal contexts play different roles and are interwoven throughout the encountering process of serendipity.

An extension of the existing framework
We identified three processes for experiencing serendipity: "unexpectedness", "connection-making" and "value". This identification is similar to the framework proposed by Makri and Blandford (2012b), in which they considered "unexpectedness", "insight" and "value" as the three key elements with which to evaluate serendipity, and where the "making of the connection itself involves an amount of insight" (p. 714). Our new contribution to this framework is that, based on our empirical study, we have further expanded the three processes. We have identified three different channels that lead to "unexpectedness", the different situations of connection-making between the encountered unexpectedness and the precipitating conditions of visceral needs, conscious needs and previous knowledge/experience, and we have also identified value as substantial value and emotional value.
The identification of "unexpectedness" in our work is similar to the work by Foster and Ford (2003), in which they classified four different categories of serendipity. Our identification of "unforeseen means of encountering information" is similar to Foster and Ford's third category, "the unexpected finding of information the existence and and/or location of which was unexpected, rather than the value" (p.332), and the identification of the "unexpected content of the encountered information" is similar to Foster and Ford's fourth category, "the unexpected finding of information that also proved to be of unexpected value: (a) by looking in "likely" sources"; (b) by chance" (p.332). However, our work also goes beyond their framework by identifying the processes of "connection-making" and "value", which is not discussed in their paper. By expanding the different processes of serendipity, we have found it is  Table 5 categorises the cases listed in the previous sections of this paper, and we argue: • For a conscious need/concern, it is more the unforeseen means of encountering the information that results in the participants' sense of unexpectedness, and brings both substantial value (e.g. an answer to the concern) and emotional value (the positive emotion when the need is addressed), as identified in Case 7, Case 44 and Case 30.
• For a visceral need/concern, it is usually the unexpected content of the encountered information that leads to participants' feelings of unexpectedness, while it also brings both substantial value (e.g. finding a possible solution for a visceral need) and emotional value (a positive emotion), as identified in Case 6 and Case 44. However, sometimes the unforeseen means of encountering information may also play a role in leading to unexpectedness, as identified in Case 1.
• For previous experience/knowledge, it often results in emotional value, and both the unforeseen means of encountering the information and the unexpected content of the encountered information have the potential to bring a feeling of unexpectedness to participants, as identified in Case 20 and Case 60.

Design strategies based on the identified contextual factors
Björneborn (2017) argues that: We cannot design environments always leading to serendipity -as serendipity is a highly subjective and situational phenomenon. But affordances for serendipity can be engineered……Serendipity may thus be intended by designers, but must always be unplanned by users (p. 1068).
From this empirical study, the identified contextual factors of external context, social context and internal context helped us to look into the role played by context during the different processes of serendipity, and thus provide possible implications for designers to design affordances that can "engineer serendipity". These include: (1) consider participants' status. Our participants tended to experience serendipity more often in the afternoon than in the morning, especially when they are in their leisure time. When organising activities that aim to facilitate serendipity (e.g. free discussion seminars, using information systems to make recommendations), it is better to take participants' status into consideration and arrange such activities during their leisure time; (2) consider locations. Our study has found that places with rich resources, such as a physical resource (e.g. licensed online libraries) or a socialisation resource (e.g. interdisciplinary offices), and with easy accessibility provide greater potential for participants to encounter serendipity; (3) try to create social networks. Social context is a significant stimulus, where participants can encounter unexpectedness which may result in serendipity, so attention should be given to such context to help to "engineer serendipity"; (4) design environments that are both diverse and conspicuous.
Motivated by an understanding of the internal context of pre-conditions and sagacity/noticeability, we also suggest the design of more diverse and noticeable environments to encourage the occurrence of serendipity. Current information technologies, such as recommendations, personalisation and visualisation may consider this aspect to facilitate the occurrence of serendipity; (5) combine emotional design. As identified in the study, emotion plays an important role during the process of encountering serendipity, so an element of curious and/or interesting information may help to encourage the occurrence of serendipity. We consider that designs for joy, surprise and/or other emotional design strategies can also be applied to serendipitous design strategies.

Two limitations of the research method
There are two methodology-based issues raised from the empirical study that we think are worthy of further discussion. The first issue concerns the participants' initial understanding of serendipity. According to our empirical studies, we have found there are disparities in the basic understanding of the term "serendipity" between the Chinese participants and the previous participants in the UK. People from non-English speaking countries, such as China, may lack an initial understanding of the concept of serendipity because the concept originated in the West. Even the Chinese translation of the term was seen as exotic by the Chinese participants. Hence, at the very beginning of the study, when we introduced the research purpose to our participants, in addition to giving the Oxford Concise English Dictionary definition of serendipity, we also used an example from the pilot study to help with their understanding of this subjective term. This undoubtedly raised the concern that participants may have restricted the types of examples they identified for experiencing serendipity during the following study. Although we repeatedly highlighted the fact that this was only one example of a personal case for encountering serendipity, and that they could send us details of any encounter during the following two weeks that conformed to the dictionary definition of serendipity, we are not fully confident that every participant was not restricted by the given example. Indeed, two participants only sent us one case during the whole study. Nevertheless, we are confident that the majority of the participants in this study were not restricted by the example, as evident in the different categories of serendipity identified from the study. We hope this concern will provide guidance to future researchers when conducting similar studies, especially in the case of cross-cultural groups lacking an initial understanding of serendipity.
The second concern relates to the use of the daily reminder during the study, which may have pressurised the participants into responding. Each reminder was sent at approximately 10:30pm, based on feedback from the pilot study and observation of most participants' routines, as this was the time when they had finished their daily work or study and were in a relatively leisurely state, and when serendipity is more likely to occur (Sun et al., 2011). During the post-interview, a few participants reported that they felt pressurised into responding on receipt of the reminder in the first two days of the study. This was mainly because it was the first time they had heard the word "serendipity", and they were not sure whether they would be able to recall any serendipitous moments in their daily activities. However, as the diary studies continued, any pressure caused by the reminder became much less obvious because they found serendipity was not such a cryptic phenomenon. They tended to understand from the reminders that there were other participants who had already experienced serendipity. This, in turn, gave them the confidence and assurance to reflect on their encounters during the day. By way of comparison, there were four participants who explained that they were particularly in favour of the reminder, and they even argued that it would have been better to send reminders twice a day (in the morning and the evening) to better remind them they were in a study situation. We hope future studies will pay special attention to this concern if employing similar

Conclusion
This paper presents a follow-up to an empirical study which was conducted in the UK in 2011. We employed the social media platform "Wechat" as a research tool to investigate the phenomenon and occurrence of serendipity in the context of information research among a group of Chinese scholars.
Based on the collected data, we have found that current understandings of serendipity, which have been constructed mainly on the basis of Westerners, also applied to our Chinese participants. They also experienced serendipity according to the three main processes of encountering unexpectedness, connection-making and recognising the value. In addition, we further expanded the definition of the three processes. Unexpectedness is encountered by any unforeseen means and/or the unexpected content of information encountering, where connections are made between the unexpectedness and the precipitating conditions of visceral needs, conscious needs, or previous experience/knowledge. Ultimately, either a substantial value or an emotional value prompts the individuals' recognition of serendipity. This expanded framework also helped us to classify the categories of serendipity.
The role of context in experiencing serendipity has been further investigated. Compared to the model present in our previous study, the updated context-based serendipity model better demonstrates the different interactions and influences of the external context, social context and internal context during the different processes of serendipity. In particular, our study found that the role of emotions should not be considered only as an outcome of serendipity, but it should also be embedded in the process of encountering serendipity, which is an issue that has been largely neglected in existing studies.
Future research should pay particular attention to the methodological issues when conducting serendipity studies cross-culturally, especially within participant groups who lack an initial understanding of serendipity. Deliberate consideration of how to introduce the term "serendipity" to participants should be undertaken before any study. In addition, the research findings relating to the differences between the UK scholars and Chinese scholars in terms of "power-distance", which exists in the social   (2006). "Psychological factors behind incidental information acquisition", Library and Information Science Research, Vol. 28 No. 4, pp. 579-594. doi: 10.1016/j.lisr.2006.03.022 Hofstede, Geert. (1980. "Culture and organizations", International Studies of Management & Organization, Vol. 10 No. 4, pp.15-41. Kefalidou, Genovefa, & Sharples, Sarah. (2016). "Encouraging serendipity in research: Designing technologies to support connection-making",   Talja, Sanna. (1997). "Constituting 'information' and 'user' as research objects: a theory of knowledge formations as an alternative to the information mantheory", Information seeking in context, pp. 67-80. Talja, Sanna, Keso, Heidi, & Pietiläinen, Tarja. (1999). "The production of 'context'in information seeking research: a metatheoretical view", Information Processing & Management, Vol. 35 No. 6, pp.751-763. Talja, Sanna, Tuominen, Kimmo, & Savolainen, Reijo. (2005). " 'Isms' in information science: constructivism, collectivism and constructionism", Journal of documentation, Vol. 61 No. 1, pp. 79-101. Taylor, Robert S. (2015). "Question-Negotiation and Information Seeking in Libraries", College & Research Libraries, Vol. 76 No. 3, pp. 251-267. Wilson, T. D. (1981). "On user studies and information needs", Journal of Documentation,Vol. 37 No. 1,.  (c) daily reminder sent to participants.           "I have received eleven messages today, ten days left for the study" indicates that the researchers have successfully received eleven serendipity messages from all the participants on that day, and there are ten days left before the study finishes. (1) What is your understanding of serendipity now that the study has concluded?
(2) When comparing your current understanding of serendipity to your understanding before the study, do you think are there any differences? If yes, why?

Part 2: For a Detailed Serendipity Case
(1) When did the case happen?
(2) Where did the serendipity happen?
(3) Could you describe the case in more detail?
(4) Why would you consider it as serendipity?
(5) What was your socialisation context in this case?
(6) How did you deal with the serendipitous information? What did you do when you encountered the serendipity?
(7) What was your emotion after you encountered the serendipity?

Part 3: Questions about the Research Method Employed
(1) What do you think about the reminder information that I sent to you every day?
(2) Do you have any suggestions or opinions on such reminders?
(3) Why did you send your case in text (or picture)?
(4) What do you think about the designed interface of the application?
(5) What was your experience of the study? Do you have any suggestions or opinions about the study or the research method?
*Note: The language used across the study was Chinese, including all the contents presented in the appendices. We have translated everything into English for this paper. Thank you for your thoughts and efforts devoted to our paper. Your comments have helped us to substantially improve the quality of our paper. We have made significant changes to our manuscript based on your comments. Changes have been made in the updated manuscript and they are briefly described in the following (for those quoted references, please refer to the manuscript):

Reviewers Comments to Author:
This is interesting work, but not yet ready for publication. The research aims have now been revised to: 1) identify whether current understandings of serendipity can also be adapted to Chinese scholars; 2) further investigate the role context plays during the different processes of experiencing serendipity. Two main changes have been made in this revised version, compared to the original.
First, the original research aim of "investigating culture differences" has been removed. After deliberation of the reviewer's suggestion (see the 20 th review comments), we agree that it is appropriate to remove from this paper the discussions about culture, and to be more focused on the issues relating to context.
The second change is that throughout the paper we are no longer using the term "employing Makri and Blandford's framework". This is because we began our coding of the collected data based on our original defined themes of understanding serendipity from our previous study, and when we identified internal contextual factors from this empirical study, we further determined the themes to evaluate participants' experiencing of serendipity as "unexpectedness" "connection-making" and "value". This framework is consistent with Makri and Blandford's framework (2012b) of evaluating serendipity with the three elements of "unexpectedness" "insight" and "value", and they argued that "making of the connection itself involves an amount of insight" (p. 714). In our original version, due to the limitation on the length of the paper (as we needed to discuss both culture and context), we didn't extend the description of our coding work, and just considered that it conformed to the existing Makri and Blandford framework. Now that we have removed the culture part, and have fully focused on the context throughout the paper, we have introduced our coding process in detail in Section 3.4 "Data analysis". We also had a discussion about both our framework and Makri and Blanford's framework in Section 6.3.1, where we formed the opinion that our contribution lies in a further extension of this framework into detailed sub-themes, namely: three different channels that lead to "unexpectedness", the different situations of connection-making between the encountered unexpectedness and the precipitating conditions of visceral need, conscious need and previous knowledge/experience, and we also identified the values as substantial value and emotional value. The expanded sub-themes make it possible to classify the categories of serendipity. (Please refer to Section 6.3.1 for further details) The term "specifically defined context" has been removed, and our aim is to investigate the role of context during each process of serendipitous encountering, based on our own contextual factors identified from the empirical study.

More details on the previous UK diary study are needed in order to understand how your Chinese findings compare to it.
This part has been revised from two perspectives. First, during the "introduction", we have demonstrated the model from a previous UK diary study, and also explained our previous research findings. The second perspective is that we have added Section 6.2 in the discussion of the revised paper, which provides a detailed comparison between the proposed new model and previous model. We have examined Belkin's ASK theory and the related chapter in the book by Case and Given, and compared to Belkin's ASK theory, we are more motivated in Taylor's work on information needs, and have added this part to Section 4.2. We found this work provides greater benefits for our identification of the precipitating conditions of internal contexts. In particular, we have found that Taylor's description of "visceral need" is more appropriate to the identified situation of our participants, and we have changed the original description of "unconscious need" into "visceral need".

To better articulate the originality/value of the paper, explain how your
context model 'provides a further understanding of the role of context...' How exactly does it enrich our current understanding of serendipity?
In the Introduction, we have now explained that, limited by the perception of serendipity at that time, our previous study only discussed these elements as different factors having an influence on participants' readiness to experience serendipity, but failed to have a further examination of how these elements would act during the separated processes of serendipitous encountering, and this is now considered as a research objective.
In terms of the originality/value of the paper, we have explained this in detail in the revised "discussion" section. Section 6.2 presents the difference between the proposed updated model in this paper and the previous model. Section 6.3 demonstrates two implications of the updated model. Section 6.3.1 presents an extension to the existing framework of serendipity, which can be used to classify different categories of serendipity cases, and Section 6.3.2 proposes different design strategies based on the new contextual factors identified from this empirical study. After deliberation of the reviewer's suggestion (see the 20 th review comments), we agree that it is appropriate to remove the discussions of culture from this paper, and to focus more on the issues of context. Due to the limited number of participants, it is no possible for us to make strong comparisons between the scholars in the UK and China.

What cultural factors
However, we do think that this is still missing from current serendipity studies. Yeh's (2007) research found culture can impact individual's information behaviours, and current research on Culture Neuroscience also provides evidence that individuals with different cultural backgrounds can perform differently in the psychological processes.
Our empirical studies also found that the role of social context can be impacted by power-distance, a widely recognised cultural factor in respect of differences between UK and Chinese scholars. While serendipity is a cryptic phenomenon which is actually highly related to an individual's psychological processes (e.g. emotion, attention, memory), we believe that it is worth investigating further from a cultural perspective. Our future research will design more strict comparative studies and will examine whether culture plays a role in serendipitous encounters. This part is now revised in the "Introduction" section. We have presented our previous context model of serendipity, and have also introduced the research findings from the previous study, which only discussed context as a factor to influence participants' readiness to experience serendipity, and failed to make a further examination of how these elements would act during the separated processes of serendipitous encountering. This is now considered as a research objective and will be further investigated in this paper. We have made the appropriate revisions in Section 2.1, giving a different researcher's definition of serendipity, and the importance of context in studying serendipity has been highlighted by citing Case and Given's (2016) argument concerning the significant role of context in information seeking. This is followed with evidence from existing serendipity studies that demonstrates the important role of context in experiencing serendipity. For example, the empirical study by Points et al. (2015) found that contextual factors such as location, activity and focus can influence a user's experience of serendipity. Similar findings can be found in another paper After a review of Erdelez's research paper, we have now changed the term "incidental information encountering" to "information encountering" and "opportunistic acquisition of information". A more detailed discussion relating to each module is included at the end of Section 2.3, especially how contextual factors are involved in these models. For example, we discussed "notice" as an internal context in Erdelez's model, "trigger" as an external context, and "precipitate condition" as both internal and social contexts in McCay-Peet's (2010) model. "Trigger" is discussed as an external context, "openness", "prepared mind" and "ability to make connections" as internal contexts in McCay-Peet's (2015) updated model, and "unexpected event" as an external context, "prepared mind" as an internal context in Lawley and Tompkins' (2008) model. Finally, we discussed "unexpected circumstances" as an external context, "insight", "mood or feelings" as internal contexts in Makri and Blandford's (2012) model, and "chance" as an external context, and "prepared mind", "act of noticing" and "surprise" as internal contexts in Rubin et al.'s (2011) model. However, a systematic discussion from the perspective of these contextual factors cannot be drawn from these existing studies.
We are also no longer using the term "employing Makri and Blandford's framework" throughout the paper. As explained in the answer to the first review comments, we achieved a similar framework from this new empirical study to evaluate participants' experiencing of serendipity as "unexpectedness", "connection-making" and "value".
This framework is similar to Makri and Blandford's framework (2012b) for evaluating serendipity by the three elements of "unexpectedness", "insight" and value", and they have argued that "making of the connection itself involves an amount of insight" (p. 714). In our original version, owing to the limitation of the length of the paper (as we needed to discuss both culture and context), we didn't extend the description of our coding work, and simply considered it as conforming to the existing Makri and Blandford framework. Now that we have removed the element of culture and have focused fully on the context throughout the paper, we have introduced our coding process in detail in Section 3.4 "Data analysis". We have also included a discussion of our framework with Makri and Blanford's framework in Section 6.3.1, where we form the opinion that our contribution lies in a further extension of this framework into the detailed sub-themes of: three different channels that lead to "unexpectedness", the different situations of connection-making between the encountered unexpectedness and the precipitating conditions of visceral need, conscious need and previous knowledge/experience, and we also identified values as substantial value and emotional value. The expanded sub-themes make it possible to classify the categories of serendipity. (Please refer to Section 6.3.1 for further details)

It would be useful to integrate McCay-Peet's (2015) model revisions into your
earlier discussion of her model.
McCay-Peet's updated model has been integrated after her early model, and a discussion of the contextual factors of both models is also introduced at the end of Section 2.3, where we have argued that her early model identified "precipitating conditions" as "active learning" (internal context) and "social networks" (social context), and the requirement for a "trigger" (e.g. text, images, audio) to facilitate serendipity. In the updated model, she further proposed different external factors (trigger-richness, highlighting triggers, enabling connections and enabling capturing) and internal factors (openness, prepared minds, the ability to make connections) that may influence the perception of serendipity, but these factors, especially the external factors, are not discussed from the perspective of context.  At the very beginning of Section 3 "Research Method", we have now explained the reason for using Wechat as the platform, and we argue there are four reasons:1) it covers similar functions to our previously designed diary application in which different types of data can be recorded and transferred (i.e. text, video, audio, and image); 2) participants can use their own mobile phone and now have no concerns about portability problems of extra devices, and no additional package needed to be installed on th e mobile phones, as they were all frequent users of "Wechat", and were quite familiar with its functions; 3) "Wechat" is a social media platform, thus it also has the function of communication between participants and researchers, so when participants had any problems during the experiment period, they could send messages to the researchers and receive responses instantly; 4) it also has the advantage of allowing the researchers to send "reminders" to participants every day, to help keep them aware that they are in an experiment situation.

It would also be useful to discuss Foster and
As the element of culture has been removed from this paper, the cultural differences were not discussed on this issue.
12. Why did you restrict your study to PhD students? How can you be sure they are representative of the broader population? Was there a special reason you recruited them? Explain in detail. There are two main reasons for recruiting PhD students: 1) following the research findings of Foster and Ford (2003), which showed that serendipity is experienced widely among researchers, we decided that PhD students were an appropriate group of scholars dedicated to research projects who were easy to access; 2) our previous study recruited 11 PhD students and received 23 serendipity cases within a week. This successful experience demonstrated that it was a feasible solution to recruit PhD students with which to conduct such a diary-study. As a result, we recruited PhD students with research experience (a minimum of 12 months) to our study. We have also made this revision in Section 3.1.

You mention that it was surprising that students did not know about
serendipity beforehand (p.10), but as you mention in your discussion, this concept is rarely used in Chinese cultures. Therefore this may not be too surprising.
In the beginning, we felt it was surprising because we found the Chinese scholars had no conception of serendipity, and most of them heard about this term for the first time.
This was quite different from our previous study on UK scholars, who all had some understanding of the concept. We were unaware of this situation at the beginning of the study. This issue led directly to a methodology based concern: how can we better introduce "serendipity" to these participants? We then addressed this concern by giving our participants the dictionary definition, in addition to an example from the pilot study. This concern is now discussed in Section 6.4.
In this revised version, as we have removed the discussions about culture, we have also removed the term "to our surprise". 14. Providing an example from your pilot can potentially help participants understand the concept of serendipity, but might also restrict the types of examples they provide in the diary study. Did you mitigate for this? If so, how?
Again, explaining this would be useful.
During the study, we explained to our participants that this was merely one case from the particular participant. We also clarified that there are different types of serendipity, and that they could send messages to us if they thought they were experiencing We have added Section 6.4 in the discussion, which discusses the limitations of the study approach. We have considered this problem as a limitation of the research methodology and hope that future studies will address this concern, especially when conducting cross-culture studies with groups lacking an initial understanding of serendipity (similar to our Chinese participants).
15. 30 mins. seems a long time to collect understandings of serendipity and introduce to Wechat. Did anything else happen during interviews? What instructions were participants given of what information to capture and how?
Were they restricted to capturing only information on the mobile Web, or any information they encountered?
We have added the following sentence in Section 3.2: "The research purpose was introduced to them, and participants were invited to collect any cases they considered to be serendipity in the following two weeks, either on the Web or surrounding their daily activities (e.g. reading, research, and socialisation)." Based on Figure 2, presented in the paper, participants were instructed to use different forms of input (e.g. text, voice, video and graphs) to send messages to the researchers, with each message including the following information for the experience: time, location, activity, emotion and impact, which is the background interface in Figure 2. A detailed instruction of this figure has been added to appendix A.
16. What exactly did you ask during the post-study interview and why? Much more explanation and justification is needed.
What are examples codes you created, merged, subsumed etc.? Your method section is lacking in specific detail.
After a careful review of our coding schemes, we consider we have employed thematic analysis to deal with our coding. We first followed a top-bottom theoretical thematic analysis to investigate how the Chinese participants experienced serendipity.
We began this part of the coding by identifying the themes drawn from our previous study, where we identified the nature of serendipity with two different levels of abstraction, and the value of serendipity. The first level identified the "unexpected finding of information" by considering different combinations of three components: whether the information is directly related to the activity being undertaken by the individual (non-activity-based vs. activity-based); whether or not the information encountered is unexpectedly valuable to the individuals (unexpectedly valuable or not); and whether the information is from an unexpected or likely source. The second level identified the making of unexpected connections between different pieces of information, people and ideas.
We then used a bottom-top inductive thematic analysis to identify the contextual factors present in the cases of serendipity. Initially, we identified a number of categories, including the time for experiencing serendipity (i.e., a.m., p.m., and across time periods), different locations when serendipity occurs (e.g. office, dormitory, classroom, library, etc.), different activities during which serendipity occurs (e.g. travelling, surfing the Internet, attending seminars, talking to classmates, talking to friends, etc.), and a category more related to an individual's cognitive or psychological characteristics, such as memories, insight, expertise, previous needs, instantly raised needs, and emotions (see Table 3 for an example of the coding for the pilot study case). We then compared the categories and grouped those with overlapping meanings into possible themes. As a result, we ended this layer of analysis with the three major themes of external context (i.e. time, location, and personal status), social context (e.g. different social counterparts) and internal context After we finished coding the themes of contexts, especially the internal context themes of the precipitating conditions, we reviewed all our coded themes and further found that our original coding of the first level "unexpected finding of information" can also be considered as a process of making connections between the encountering and the precipitating conditions. As a result, we re-coded this part of the framework into the three different themes of unexpectedness, connection-making and value, and these were further expanded into sub-themes, namely: "unexpectedness" into the themes of "unforeseen means of encountering information", "unexpected content of the encountered information" and "both"; "connection-making" was further expanded into "connection-making between unexpectedness and visceral need", "connectionmaking between unexpectedness and conscious need" and "connection-making between unexpectedness and previous experience/knowledge"; and "value" was expanded into sub-themes of "substantial value" and "emotional value". This is also why in our original version we applied Makri and Blandford's framework in identifying our coding, as our re-coded themes of "unexpectedness" "connectionmaking" and "value" are quite similar to those used in their framework. However, after a clear introduction of this element in the revised paper, we have no longer used the term "employing Makri and Blandford's framework".
This has been revised accordingly in Section 3.4 "Data analysis", where we have also added a new Table 3 to explain our coding scheme.
The work of Makri and Blandford has been reviewed and evaluated. In fact, it is their framework (2012b) rather than their model (2012a) which is employed in our original paper. As explained in the 17 th reply, we have now shown that the framework is drawn from our own coding on the collected data, so we no longer use "employing  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  We have supplemented the case of the fire extinguisher, and also go through all the participants' cases used in this paper to ensure validity of our arguments. We now believe every case has an instruction to its relevant discussed topics.
In the case of the fire extinguisher, the participant reported that she had learnt about the principles of a fire extinguisher from a talk during a training session. She then raised a need to collect relevant information about fire extinguishers around her lab setting (e.g. where they are located in the lab, how to use them). However, this need was not addressed at that moment and the participant forgot to address the need when the talk was finished. It was not until she accidently attended a related "fire alarm introduction" that she remembered the need, and found the answer to this need during the presentation. We argue that the answer to the participant's need was not unexpected to her, but the unforeseen means by which she received the answer induced feelings of "unexpectedness", as attending such a presentation was not on her original schedule. This has been revised accordingly in Section 4.1.
With regard to the use of the term "super-encounterer", it is not because of the participant's sagacity on the three different occasions, but from the collected serendipity cases from all participants during the two-week study. Some participants sent us 13 cases during the study, while others only sent one case. We argue that such differences reflect the differences in each participant's sagacity, and those who sent 13 cases may be considered as "super-encounterers" (Erdelez, 1997). This paper has been revised significantly. The culture section of the original paper has been removed, and the focus is now on the issues of context.

It's not clear what 'measure by which information is encountered' (p.14)
means and this makes it difficult to understand this part of your findings (and their importance).
The original sentence has been revised to "Both the unforeseen means and content of the encountered information bring a sense of unexpectedness", to keep it in accordance with the prior descriptions in Section 4.1.

It is worth referring to Taylor's (1964) seminal paper conscious and unconscious needs in 4.2.
Taylor's work has been reviewed carefully. In particular, we have found our previous definition of "unconscious need" is more accurate as "visceral need", which was referred to by Taylor as "not existing in the remembered experience of the inquirer", who also argued that it probably "is inexpressible in linguistic terms", and can "change in form, quality, concreteness, and criteria as information is added". We have explained this in Section 4.2.

Can't an example represent both types of value (i.e. emotional too?)
In Sections 5.3.3 and 6.3, we have now pointed out that both emotional value and substantial value will exist in any "need-oriented" serendipity cases, as either finding the answer to previous concerns (conscious need-oriented cases) or finding a possible solution to the raised visceral need may lead to a positive emotion for the participant  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  We have now explained that the contextual factors of time, location and personal status are identified based on our own coding of the collect data, and Section 5 "Role of Context" has been largely revised. We have now explained the reasons for these contextual factors affecting how people experience serendipity, as follows: Time: Existing research has demonstrated that different times of day can impact human performance (Fröberg, 1977) and even the cognitive and evaluative efficiency of individuals (Natale et al., 2003). Our participants also reported that they were more engaged in different activities in the afternoon, as a result of which it was also more likely that they would encounter serendipity.
Location: We investigated the reasons why locations such as an office environment would produce the most cases of serendipity and found that: 1) these places are source-rich; 2) there is an interdisciplinary social setting; and 3) participants could readily gain access to these resources.
Personal status: participants in a more open and relaxed state (e.g. during leisure time) encounter serendipity more frequently when compared to other statuses, such as when attending a seminar or working/studying.
We also added a possible reason for social context, where participants are found to experience serendipity more often when they are communicating with their peers.
When compared to our previous findings from the UK study -where four of the seven social context related cases happened during participants' communication with their superiors -we think the cultural difference known as "power-distance" played a role in this situation. 26. While it clear that your findings support those from several previous works, you do not make a strong case for why your work is novel and important. This is essential in any empirical research.

Include page numbers when quoting directly from an external source.
Substantial revisions have been made in the discussion section, and compared to the previous model, the contribution of this new model lies in the following: 1) It helps to explain the different roles of external context, social context and internal context during each process of experiencing serendipity (unexpectedness, connectionmaking and value), which provides evidence that serendipity is the "product of context" (Foster and Ellis, 2014, P. 18). This is discussed in detail in Section 6.2.
2) Based on the identified internal contextual factors of precipitating conditions, we expand the framework of "unexpectedness", "connection-making" and "value" by further identifying the sources of unexpectedness (unforeseen method of encountering information and/or the unexpected content of such information), connection-making between unexpectedness and precipitating conditions (i.e. visceral needs, conscious needs and previous experience/knowledge), and value (substantial value and emotional value). Such expansion made it possible for us to classify the categories of serendipity. See a detailed discussion of this in Section 6.3.1.
3) Based on the contextual factors identified, design implications can be drawn to "engineer serendipity", such as considering participants' status, considering locations, trying to create social networks, designing environments that are diverse and noticeable, and combining emotional design. See Section 6.3.2.
Discussion 27.You present some thoughtful discussion. However, it would be useful to discuss your findings in light of the existing literature -particularly on contextmore. Much of your discussion section reads like a findings section.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  The discussion section has been largely restructured. We have now removed references to culture and have focused on issues of context. Sections 6.2 and 6.3 have been added to demonstrate the novelty of our research findings. After a discussion of the updated model in Section 6.1, in Section 6.2, we have compared this model with the previous model, and extend the current literature by unpacking how these contextual factors affect the different processes of a serendipitous episode. Section 6.3.1 discusses the new classification of the categories of serendipity, by expanding the identified framework from the empirical study, while in Section 6.3.2, following Björneborn's (2017, p. 1068) argument that "We cannot design environments always leading to serendipity…… but affordances for serendipity can be engineered", and based on the identified external context, social context and internal context from this empirical study, we propose different implications for designers to include affordances that can "engineer serendipity". The discussion of the differences in culture between the UK and Chinese scholars has been removed from the paper, and it is now focused on issues of context. We will make more strict comparisons between different cultural groups in future studies. We have added Section 6.4 for the discussion of the limitations of the study approach, in which two concerns are discussed. The first is the participants' initial understanding of serendipity. In addition to the dictionary definition, the lack of initial understanding from the Chinese participants prompted us use an example from the pilot study to help explain this new term. This raised the concern that participants may restrict the types of examples for experiencing serendipity during the following study.
The second limitation is the setting of the daily reminder. A two-week study is a relatively long time, and sending reminders to participants was intended to ensure they remembered that they were in an experiment situation, although it may have caused additional pressure to be put on participants. Thus, how to balance such reminders is another concern that needs to be considered carefully in any future research.