The Research Design

The challenge of a research design for investigating serendipity in a controlled environment is to ensure that participants have a serendipitous experience during a research session. We developed an experimental website and recruited 260 participants to perform search tasks. The website applies real data from a Chinese online reading community. It contains three starting pages, each page presents:

• A foreground task: Your best friend’s birthday is coming up. You need to find books as the gift for your friend and add them to the gift list;
• Tags in one of the three different presentation formats: list, cloud, and tree;
• A background task: Add any serendipitous books useful for your research to the favorite list, and
• Related links for participants to take actions.

In this study, serendipity was conceptualized as a participant’s transition from seeking resources relevant to the foreground task to serendipitously discovering useful resources relevant to the background task. One of the independent variables is tag presentation format. Three tag formats were designed:

• Tag list simply lists all tags by the number of books attached. All tags are of the same font size, color and orientation.
• Tag cloud applies different font sizes, colors and orientations to reflect the importance of tags. The more books attached, the more important a tag becomes and the larger its font.
• Tag tree, with a 3- or 4- layer tree structure, was constructed through a Card Sorting survey. The font color and unit of indention give a visual cue of which tags are related to each other hierarchically.

The other independent variable of this study is openness to experience. It is a personality trait that describes people who are intellectual, creative, unconventional and innovative. Its relationship with serendipity has been inconsistent in previous research.

To test the relationship between perceived serendipity and tag presentation formats and that between perceived serendipity and openness to experience, participants were asked to take a re-experiment online survey after a brief introduction of the research, then conduct a book search to complete the foreground and background tasks, and take a post-experiment online survey at the end. User data such as tag presentation forma selected, time spent on survey and searching tasks, books selected were also recorded. At last, results from 207 participants were used for hypothesis testing and data analysis.

Results and Significance

We found that tag clouds and trees attract users’ attention to potential useful resources by highlighting visual cues. And tag trees with rich semantic information help people recall relevant knowledge and make meaningful connections between resources, resulting in higher perceived serendipity. These explain why people experience serendipity more frequently while using tag trees than clouds, followed by lists. Although openness does not influence serendipity across tag formats, further analysis shows that it significantly decreases serendipity for lists, but significantly increases serendipity for clouds and trees. The reason is that more open people have higher aesthetic pursuits and yearn for aesthetic information technology artifacts. They prefer using pleasing tag formats, namely clouds and trees. Whereas less open people care less about the aesthetic of tag formats, and prefer using traditional and familiar lists for information search activities. When people use tag formats that match their preferences, they are more enjoyable exploring communities and have more chances to gain unexpected but valuable information.

Our study contributes to serendipity research. It reveals that perceived serendipity can be influenced by interface feature of communities (tag presentation formats) and individual difference (openness to experience). Tags, users’ keyword-based explanations to resources, are regarded as triggers to serendipity. We suggest that different ways for communities to visually present tags influence user perceived serendipity. Openness to experience is a personality trait that describes people who are intellectual, creative, unconventional and innovative. Its relationship with serendipity has been inconsistent in previous research. This study seems to confirm findings that openness does not influence perceived serendipity. We actually go one step further to reconcile the seemingly contradictory findings by demonstrating an interesting interaction effect. That is, the impact of openness on perceived serendipity depends on tag formats.

The practical implication of this study is that for online communities, the provision of appropriate tag formats is important to cultivate serendipity. Tag trees brings about the highest serendipity, but they only apply to communities where there are hierarchical relationships between tags. Thus, we recommend that communities apply ontology technology or build user-generated mechanisms to help establish semantic relationships between tags. Clouds are recommended for communities that have not yet established tag relationships, as clouds facilitate serendipity better than lists by highlighting tags as triggers. More generally, our findings provide evidence that serendipity can be “designed” or “engineered” through interface features. Approaching serendipity at the feature level provides greater possibility for the design of serendipitous online communities, which improves the efficiency of knowledge acquisition and discovery.

Future research into serendipity in communities should explore it under other factors, such as more diverse categories of community services, various task types and information activities. Our future research includes exploring how different task types (e.g., work-related, recreation-related) influence perceived serendipity, and studying serendipity involved with other information activities (e.g., browsing without any search targets). And more work is warranted to understand how tag variations (e.g., visual, textual or both) impact perceived serendipity. These investigations would help the online communities to better design or revise their platforms.

The original article on which this essay is based is: Qin, C., Liu, Y., Ma, X., Chen, J., & Liang, H. (2022). Designing for serendipity in online knowledge communities: An investigation of tag presentation formats and openness to experience. Journal of the Association for Information Science and Technology, 1–17. https://doi.org/10.1002/asi.24640.