TL;DR
This paper challenges the traditional view of search as a solitary activity by presenting a model of "social search." Through a survey of 150 users, the authors demonstrate that social interactions significantly influence information seeking before, during, and after the search process, particularly for informational searches. This work suggests that future search systems should better support these social dynamics.
What Problem Does This Paper Address?
The prevailing paradigm in search engine research has largely treated information seeking as an individual endeavor. While some work has acknowledged collaborative search, the authors argue that this overlooks the broader spectrum of social interactions that can impact a user's search experience. They aim to address this gap by proposing a more comprehensive model that integrates social elements into the entire search lifecycle, from defining information needs to acting upon search results. This is important because understanding these social influences can lead to the design of more effective and user-centric search tools.
Methods
The authors conducted a critical-incident survey with 150 users recruited via Amazon's Mechanical Turk. This method involved prompting users to recall and describe their most recent search experiences. The survey questions focused on the context and purpose of the search, as well as any social interactions that occurred before, during, and after the search itself. The collected data was then used to inform and integrate existing models of sensemaking and information seeking into a novel "social search" model.
Datasets
The study utilized a dataset of 150 user-reported critical incidents of search behavior collected through a survey on Amazon's Mechanical Turk. No pre-existing datasets were used for training or evaluation.
Key Insights and Ideas
- Search is Inherently Social: Contrary to the solitary model, social interactions are not just an add-on but an integral part of the search process for many users.
- Social Influence Across the Entire Search Lifecycle: Social interactions are valuable before (context framing, requirement refinement), during (informational search, sensemaking), and after (organization, distribution) the active search phase.
- Informational Search Benefits Most: While transactional and navigational searches may see limited social benefit during the search itself, informational searches, which are more exploratory, are significantly augmented by social input.
- Implicit Social Sharing Matters: Beyond explicit collaboration, implicitly shared information (e.g., seeing what others have searched for or shared) can also be valuable.
- Sensemaking is a Socially-Augmented Process: The process of making sense of information, especially in complex informational searches, can be greatly enhanced by discussions and feedback from others.
Key Results
- Pre-Search Social Influence: 42.0% of users refined their search requirements through social interactions, seeking advice, keywords, or guidelines.
- Informational Search Dominance: 59.3% of the reported search acts were informational in nature.
- Post-Search Social Engagement: Nearly half of users (47.3%) organized their search results, and over half (58.7%) distributed their findings to others, often for feedback or to share knowledge.
- Self-Motivated Searches are Social: For self-initiated informational searches, social interactions were prevalent both before and after the search.
Limitations and Future Work
The authors acknowledge that their survey focused on critical incidents, which might not capture all types of search behavior. They also note that their survey did not explicitly ask users to report on social incidents or collaborative search acts, meaning the reported social interactions might be an undercount.
Future work suggested includes exploring specific techniques to support social search, such as integrating instant messaging with search interfaces, leveraging community features on websites to connect users with domain experts, and displaying related search terms or automatically generated tag clouds to provide implicit social feedback. The authors encourage further research into both explicit and implicit social interactions to augment web-based search.
Summary updated