2.1 Evaluating knowledge construction
Information accumulation is a key criterion of the success of the knowledge construction process on Q&A sites [7]. Knowledge construction on Q&A sites creates new ideas, explanations, and theories that help the members of a community understand the world [8–10]. According to Latour and Woolgar [11], information is accumulated in the knowledge construction process through the negotiation of new content among knowledge contributors. On Q&A sites, the provision of informative content follows an ordered progression of factual construction over time [5, 12].
However, information accumulation has been widely ignored in evaluations of knowledge construction on Q&A sites, especially Chinese Q&A sites. Studies have examined the information listed on Q&A sites by employing measures such as accuracy; completeness; article length; information richness; and the number of references, headings, and functional links [12], but few of them have employed the criterion of information accumulation. Therefore, the first objective of this study was to use information accumulation as an objective measure to examine the process of online knowledge construction in the context of Chinese Q&A sites, taking Zhihu as a case study.
2.2 Marketplace of ideas
The marketplace of ideas, a concept underlying the notion of freedom of expression, is based on an analogy with the economic concept of a free market. With regard to Q&A sites, the marketplace of ideas refers to the act of discussion among distributed contributors. Q&A sites have become an ideal manifestation of the marketplace of ideas. Q&A sites embrace the tenets of marketplace theory, namely (1) that ideas should be in competition; (2) that the public at large is the best provider of ideas; and (3) that exposure to contradictory points of view has value in the search for truth [13]. On Q&A sites, users compete with one another to dominate the discussion, demonstrating the competition of ideas on the open market, which is the key component of marketplace theory. Q&A sites also reduce barriers to entry in the marketplace of ideas because all opinions can be freely expressed, with the open exchange of ideas. Furthermore, an essential function of Q&A sites is the ability of users to read and reply multiple answers to a question and be exposed to conflicting points of view; this can promote the diversity of opinions. These characteristics make Q&A sites ideal online communities to which marketplace theory can be applied to study knowledge construction [13].
The second objective of this study was to investigate the contribution of the marketplace of ideas to information accumulation on Q&A sites. Q&A sites can be generally classified as a competitive marketplace of ideas or a monopolistic marketplace of ideas [14]. In a competitive marketplace of ideas, the sharing of contributor viewpoints can be understood as an exercise undertaken to resolve differences of opinion [5]. Answers on Q&A sites reveal competition among ideas. In addition, in a more competitive marketplace of ideas, free expression is self-correcting [15]. A consensus emerges from competition among ideas in a free, transparent public discourse, and ideas and ideologies are winnowed on the basis of their superiority or inferiority and their acceptance by the user population.
By contrast, in a monopolistic marketplace of ideas, the flow of information is controlled. The dominant viewpoints monopolizes the marketplace and drives the other information out, thus excluding rival opinions [16]. In a Q&A site that operates as a monopolistic marketplace of ideas, the critical thinking of knowledge contributors is suppressed. Therefore, in a monopolistic marketplace of ideas, less-diverse viewpoints are disseminated to the public, thus generating less new knowledge [17]. Therefore, we proposed the following hypothesis:
Hypothesis 1
A more competitive marketplace of ideas contributes more to information accumulation than a more monopolistic marketplace of ideas does.
2.3 Participation features
On the basis of the study by Latour and Woolgar [11] and other studies on the collaborative construction of knowledge [2, 18], the present study further included participation features, temporal features, and discourse features as explanatory variables to explain the knowledge construction process.
Q&A sites promote social participation, which is pivotal to facilitating knowledge contributions because it promotes the development of social cohesion and a sense of belonging [8–10]. According to Gunawardena et al. [2], participation features determine the key stages of knowledge construction, which are the discovery and exploration of dissonance or inconsistency among ideas, concepts, or statements; the negotiation of meaning and co-construction of knowledge; and the testing and modification of the proposed synthesis or co-construction process. On Q&A sites, participation features may affect users’ motivation to contribute knowledge and the manner in which they share their existing knowledge.
2.3.1 Presence of active users
Participation feature in current study refers to the presence of active users. In online knowledge construction, most active users make crucial contributions to information accumulation. A widely observed phenomenon in online communication is the “90–9–1 rule.” This refers to the fact that in most online communities, 90% of users never contribute but only consume content, 9% of users have a little contribution, and 1% of users account for almost all of the contributions [19–21]. Studies reported the presence of uneven contributions in online collaborative settings [22, 23]. For example, Serrano et al. [24] proposed a ratio metric to calculate the inequality of participation in Wikipedia: the quotient between the number of contributions from the top 10% contributors and the number of contributions from the other 90% users. They observe a power-law in the distribution of contributions. Movshovitz-Attias et al. highlighted the importance of the contribution of active users to the answers posted on StackOverflow [25]. Users with high numbers of contributions were usually the first responders to questions. If the questions remain unanswered, they are then answered by users with fewer number of contributions. Accordingly, we posited the following hypothesis:
Hypothesis 2
Unlike ordinary users, active users promote information accumulation.
2.4 Temporal patterns of knowledge construction on Q&A sites
Temporal features refer to the time interval between answer posts and the order of answers. Temporal features result in the construction of knowledge in a hidden and unintentional manner [26, 27]. First, online knowledge creation is not a linear function of time. A marked reduction in the interval between answer posts on a site (i.e., a spike in answers posted in a given period) can serve as a proxy for a peak in online collaboration [28]. Q&A sites are web environments in which content is stored and managed collaboratively. They are also platforms on which users can showcase their knowledge and expertise to peers or potential recruiters [29]. Surges in posting behavior on Q&A sites can promote emergent collaboration [30] in terms of improvisation [31], heedful interrelating [32], privileging expertise [33], and knowledge shaping [34]. Users may provide more responses in a short period of time under external or internal triggers [30], such as a controversial topic, the presentation of biased information, or some form of vandalism (i.e., editing deliberately intended to defeat or obstruct other contributions). In this regard, a short time interval between answer posts indicates the burstiness of knowledge construction. Burstiness in this context is characterized by users collaboratively engaging in the knowledge construction process in a short period of time, during which knowledge is accumulated. Anderson et al. [35] observed that the majority of answers on Q&A sites are posted within a day after the question is asked. In addition, the longer a question remains unanswered, the lower the likelihood is that a satisfactory answer will be provided. Therefore, we postulated the following hypothesis:
Hypothesis 3
The time interval between answer posts affects the dynamics of knowledge construction. The shorter the time interval between answer posts, the higher the level of information accumulation.
Second, information accumulation is a naturally decaying process. The amount of new information decreases along with the increase of the order of answers, because multiple users offer the same information or refer to the same source for answering a single question. The first answers tend to provide novel information, whereas answers provided at the end of the thread tend to be less informative. Anderson et al. [35] reported that on StackOverflow, the most informative answers usually appear earlier in the sequence of responses to a question. Therefore, information accumulation decreases with the order of answers because of the repeated referral to similar information. We presented the following hypothesis:
Hypothesis 4
Information accumulation is a naturally decaying process represented as a function of answer order.
2.5 Discourse features
Discourse features refer to sentiment and readability. As noted by Latour and Woolgar [11], in online knowledge construction, the discourse is subjected to meaning negotiation. Information that conveys sentiment [36] and is easy to understand is more likely to stimulate online engagement.
2.5.1 Sentiment
In this study, we employed sentiment and readability as discourse factors. Sentiment can affect users’ knowledge contributions to a platform. Q&A sites facilitate numerous types of interactions. On these online collaboration platforms, members may be driven to answer questions for conveying their sentiment appropriately through text [37]. From the perspective of interpersonal communication, individuals’ sentiment toward a topic represents the atmosphere of the discussion on that topic. Expressions of gratitude and reciprocity are essential for developing trust and empathy among users, which increases their willingness to collaborate and the acquisition of answers on Q&A sites [38]. A study by Jiao et al. determined that replies that express positive sentiment (e.g., “that’s a great summary”) can lead to further communication [39]. By contrast, replies that convey negative sentiment (e.g., “I am so confused” and “Shut up”) are more likely to end a conversation [40].
Sentiment also indicates knowledge contributors’ interest in a topic. Positive sentiment of a topic on Q&A sites correspond to greater interest in that topic [41, 42]. If a group of people have positive sentiment for a topic, they are more likely to focus on that topic and contribute to the platform [43]. Accordingly, we advanced the following hypothesis:
Hypothesis 5
The positive sentiment of an online discussion leads to the accumulation of information, whereas the negative sentiment of an online discussion reduces the accumulation of information.
2.5.2 Readability
Readability refers to the textual comprehensibility of a discourse [44]. On Q&A sites, readability represents the ease with which a piece of text can be understood [45]. Journal articles written for expert audiences and answers on Q&A sites written for nonexpert readers differ substantially in their readability [46]. Knowledge constructed by expert users with extensive domain-specific knowledge is often difficult to understand because such users assume that their readers are fellow experts who can easily comprehend abstruse scientific text [47]. Online knowledge construction on Q&A sites entails the reformulation of professional content for the non-specialist public through linguistic means, which supports the popularization of this content on the sites [48]. Knowledge constructed for general users who have little or no prior knowledge of a given topic or field must be easily understandable. Higher readability can increase the probability that the text conveys the correct information with minimal reading effort [45].
The readability of text on Q&A sites may affect the dynamics of knowledge construction on such sites [45]. The use of jargon-free language may increase the accessibility and acceptability of scientific knowledge to the non-specialist public. Readable language improves the reception of a post and attracts a greater number of general users, encouraging them to join the discussion. Therefore, we proposed the following hypothesis:
Hypothesis 6
An online discussion with high readability facilitates the accumulation of information.
On the basis of the findings of Harper et al. [49] and Kim et al. [50], we classified questions posted on Q&A sites into three types: opinion based, experience based, and fact based. The question type represents the style of response required for the question. Opinion-based questions require others’ thoughts on a topic of general interest; these questions do not have a correct answer. Experience-based questions require that respondent’s share their experiences. Fact-based questions require the provision of objective information.
In addition, we employed the number of answers to each question as an additional control variable. The number of answers indicates the popularity of a question. We controlled for the number of answers to examine the effects of the marketplace type (i.e., competitive or monopolistic) on information accumulation.