Reporting serendipity in biomedical research literature : a mixed-methods analysis
As serendipity is an unexpected, anomalous, or inconsistent observation that culminates in a valuable, positive outcome (McCay-Peet & Toms, 2018, pp. 4–6), it can be inferred that effectively supporting serendipity will result in a greater incidence of the desired positive outcomes (McCay-Peet & Toms, 2018, p. 22). In order to effectively support serendipity, however, we must first understand the overall process or experience of serendipity and the factors influencing its attainment. Currently, our understanding and models of the serendipitous experience are based almost exclusively on example collections, compilations of examples of serendipity that authors and researchers have collected as they encounter them (Gries, 2009, p. 9). Unfortunately, reliance on such collections can lead to an over-representation of more vivid and dramatic examples and possible underrepresentation of more common, but less noticeable, exemplars. By applying the principles of corpus research, which involves electronic compilation of examples in existing documents, we can alleviate this problem and obtain a more balanced and representative understanding of serendipitous experiences (Gries, 2009). This three-article dissertation describes the phenomenon of serendipity, as it is recorded in biomedical research articles indexed in the PubMed Central database, in a way that might inform the development of machine compilation systems for the support of serendipity. Within this study, serendipity is generally defined as a process or experience that begins with encountering some type of information. That information is subsequently analyzed and further pursued by an individual with related knowledge, skills, and understanding, and, finally, allows them to realize a valuable outcome. The information encounter that initiates the serendipity experience exhibits qualities of unexpectedness as well as value for the user. In this mixed method study, qualitative content analysis, supported by natural language processing, and concurrent with statistical analysis, is applied to gain a robust understanding of the phenomenon of serendipity that may reveal features of serendipitous experience useful to the development of recommender system algorithms.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.