Source-linked AI summary
Towards a Model of Understanding Social Search
Brynn M. Evans, Ed H. Chi
TL;DR
Search research often treats information seeking as solitary, leaving the role of social interactions insufficiently represented. Using a survey of 150 users and integrated sensemaking and information-seeking models, this paper presents a canonical social-search model showing social inputs before, during, and after search.
Problem
Search research has largely centered on individual users, with limited attention to how other people contribute throughout information seeking.
Method
The authors surveyed 150 users about recent searches and integrated the findings with sensemaking and information-seeking models into a canonical social-search model.
Results
Social interactions occurred before, during, and after search, with informational searches showing input from others for brainstorming, opinions, and improving search schemas.
Takeaways & Limitations
The model identifies opportunities to support social interactions for task requirements, advice, feedback, brainstorming, query selection, and knowledge sharing across the search process.
Abstract
from arXiv · showhide
Search engine researchers typically depict search as the solitary activity of an individual searcher. In contrast, results from our critical-incident survey of 150 users on Amazon's Mechanical Turk service suggest that social interactions play an important role throughout the search process. Our main contribution is that we have integrated models from previous work in sensemaking and information seeking behavior to present a canonical social model of user activities before, during, and after search, suggesting where in the search process even implicitly shared information may be valuable to individual searchers.
INTRODUCTION AND RELATED WORK · SURVEY
The paper challenges individual-centered accounts of search by framing social search as including both explicit collaboration and implicit social interactions. A critical-incident survey of 150 Mechanical Turk users examined social influences before, during, and after search.
- INTRODUCTION AND RELATED WORK: Search research has largely treated information seeking as an individual activity focused on users’ needs and result relevance.Earlier information-seeking models discussed search behaviors and motivations but overlooked other individuals’ roles.
- INTRODUCTION AND RELATED WORK: Social search includes explicit collaboration and implicit social experiences involving social or expertise networks and shared workspaces.The concept encompasses collaborative co-located search as well as broader social interactions.
- SURVEY: 150 users described their most recent search act in a critical-incident survey conducted through Amazon’s Mechanical Turk.Participants reported the incident’s context, purpose, and interactions with other individuals.
- SURVEY: The survey asked what information users sought, whether they talked with anyone beforehand, and which steps they took to find it.These questions covered the information need, pre-search discussion, and search process.
- SURVEY: The survey examined whether nearby people influenced the search and what users did immediately afterward.It specifically asked about interaction with nearby people and post-search activity.
- SURVEY: Users also reported whether and how they shared information after finding it.The survey included separate questions on post-search sharing and its form.
- SURVEY: Two-thirds of search acts occurred on the survey day, 19.3% occurred the previous day, and 17.3% occurred more than 2 days earlier.Participants also provided profession, job role, and job-expertise information.
RESULTS
The results integrate the study’s findings with sensemaking and information-seeking models to present a canonical model of social search. The model examines three phases, emphasizing social information exchange through quantitative data and anecdotal user cases.
- RESULTS: The authors integrate their findings with existing models of sensemaking and information seeking.This integration forms the basis of the paper’s main contribution.
- RESULTS: The paper presents a canonical model of social search in Figure 1.The figure appears on the next page.
- RESULTS: The model is discussed across three phases, highlighting social information exchange with quantitative data and anecdotal cases of user behavior.The evidence includes both numerical findings and actual user examples.
Before Search · Context Framing
Before search, information seeking begins with a need or motivation that frames users’ motives and information needs. In the sample, most users were self-motivated, while others searched after specific requests from bosses, customers, or clients.
- Context Framing: Information-seeking behavior is rooted in a need to find information or a motivation driving the search process.The passage characterizes this need or motivation as the basis of information seeking.
- Context Framing: Context framing defines users’ motives and information needs before the search process begins.This stage establishes why users are searching and what information they seek.
- Context Framing: Information requests may originate externally or be self-initiated.The passage distinguishes requests from outside parties from users’ own motivations.
- Before Search: 47 of 150 users (31.3%) searched following a specific request from a boss, customer, or client.These users’ searches were prompted by an external request.
- Before Search: 103/150 users (68.7%) were self-motivated to search for information.Self-motivation was the larger category in the sample.
- Before Search: The sample included 150 users whose searches reflected either specific external requests or self-motivation.The passage reports both sources of search initiation within the same sample.
Requirement Refinement · During Search
Requirement refinement follows the establishment of users’ information needs and motives, while the search stage instantiates representations through traditional information seeking and foraging. Social interactions mark requirement refinement 42.0% of the time and support need influence, resource gathering, and task-guideline development.
- Requirement Refinement: After users establish an information need and motives, they refine their search requirements.This phase involves gathering requirements and formulating schemas for effective search.
- Requirement Refinement: 42.0% of requirement-refinement cycles involved social interactions, covering 63/150 users.These interactions marked the cycle as a means to influence the information need.
- Requirement Refinement: Social interactions during requirement refinement helped users influence information needs, gather keywords or URLs, and develop task guidelines.The passage identifies these as the main purposes of social interaction in this phase.
- During Search: The search stage represents active instantiation of representations, or “encodons,” within a “data coverage” loop.Although search can be cyclical, this stage is defined by representation instantiation.
- During Search: During search, users engage in traditional information-seeking and information-foraging activities.The model connects these activities to the search stage’s representation-instantiation process.
- During Search: The paper distinguishes transactional, navigational, and informational search acts.It draws special attention to the social interactions associated with these three types of search acts.
Transactional Search · Navigational Search
Transactional searches locate sources for subsequent web-mediated activities, while navigational searches identify recognizable content in particular, often familiar locations. Although both sometimes involved pre-search interactions, neither included information exchange during search, limiting the likely benefit of socially augmented retrieval.
- Transactional Search: Transactional search locates a source where users can subsequently perform a transaction or other web-mediated activity.Examples include requesting weather, movie listings, or customer-account data after navigating to a website.
- Transactional Search: Transactional searches typically followed routine steps to reach a website and request specific information.The passage gives examples involving destination weather, movie listings, and customer-account data.
- Navigational Search: Navigational search involves actions to identify content from a particular, often familiar, location.The content is often known in advance or easily recognized when rediscovered.
- Navigational Search: A navigational search example involved using Google to find the NIH website before searching there for medical-drug information.The nurse knew exactly where the information would be.
- Transactional Search · Navigational Search: 42.1% of transactional searches and 47.6% of navigational searches involved pre-search interactions with others.These interactions occurred before the search rather than during information retrieval.
- Transactional Search · Navigational Search: Information exchange did not occur during either search type, making socially augmented search unlikely to improve or facilitate retrieval.The conclusion applies specifically to transactional and navigational information retrieval.
Informational Search
Informational search was common in the survey and often benefited from social interaction. Social exchanges supported both foraging and sensemaking by shaping search representations, identifying evidence, and evaluating whether information met users’ needs.
- Informational Search: Social search may greatly improve informational search, an exploratory process combining information foraging and sensemaking.Informational search involves seeking information that may or may not be familiar to the user.
- Foraging: During foraging, users search within a patch, skim and read sources, extract information, and sometimes seek feedback while updating search representations.A public librarian worked with her boss to deduce information needed to find the Cheetah Girls 2 soundtrack.
- Sensemaking: During sensemaking, users identify evidence files, modify search schemas and queries, and organize information for later use.An English Professor copied search results into a Word document to sort and summarize for an upcoming lecture.
- Sensemaking: Social interaction augmented sensemaking through brainstorming before search and discussion afterward about whether an API evidence file was sufficient.The example involved a programmer from Intuit collaborating with a colleague while searching for an application programming interface.
- Informational Search: 39.3% of informational searchers, or 35 of 89 individuals, had social experiences before searching, including brainstorming, assessing opinions, and improving search schemas.These interactions were not simply obligatory and helped users determine what material would be useful.
After Search
After the active search phase, users often obtain an end product that may then be organized and/or distributed for action.
- After Search: After active search, an end product is often obtained and may be acted on through organization and/or distribution.The passage characterizes organization and distribution as ways the end product may be acted on.
Organization · Distribution
Users commonly organized their search end products, and these organizational acts often supported distribution to others. Sharing remained especially important after self-initiated searches, including for verification, feedback, or perceived interest.
- Organization: 71/150 users (47.3%) organized their end products in some fashion.This organization corresponded to schematizing, in which raw evidence is organized and represented schematically.
- Organization: Users organized search results through activities such as printing and reviewing them before legal inspection.One real estate agent prepared printed results for an attorney’s legal inspection.
- Organization: A design-company president bookmarked online articles about web mashups to read later in the week.This example illustrates organization for deferred review.
- Distribution: Most organizational acts served to distribute the end product to others as a presentation or publication of a case.Distribution also occurred directly face-to-face or verbally over the phone.
- Distribution: 88/150 respondents (58.7%) shared their end products with others, regardless of the distribution mechanism.One floral designer relayed information about local spring blooming flowers to a bride-to-be.
- Distribution: 49 of the 104 self-motivated searchers (47.5%) distributed search content for verification, feedback, or perceived interest.These activities indicate that social interactions remained important after the primary search act, especially for self-initiated searches.
CONCLUSION
The model suggests that social inputs can support users before, during, and after search. It proposes several ways to facilitate social search, including personal connections, domain experts, and prior users’ search knowledge.
- CONCLUSION: Social interactions may establish search-task requirements before searching, improve schemas and keyword selections during search, and provide feedback or knowledge sharing afterward.During self-motivated informational searches, users may seek advice, feedback, and brainstorming; after searching, they may collect feedback or share knowledge gained.
- CONCLUSION: A notion of “social search” may facilitate the process of information seeking.The conclusion presents social search as an overarching approach derived from social inputs across the search process.
- CONCLUSION: Potential support includes instant messaging with personal connections, expert discovery through existing communities, and displaying successful keywords or prior search trails.These alternatives connect searchers with advice, domain-specific expertise, or knowledge from previous users.