Abstract
In this lecture, we introduce exploratory search, relate it to relevant extant research, and outline the features of exploratory search systems.
Introduction
Traditional web search is built upon a "query-response" model optimized for retrieval. While this paradigm excels at "lookup" tasks—finding a specific fact, a known website, or a clear-cut answer—it often fails to support the complexity of human information-seeking. This monograph argues that for tasks involving learning, investigation, and discovery, we must move toward a new paradigm: Exploratory Search.
Exploratory search occurs when a user’s goals are ill-defined and evolve through interaction with the system. Whether a user is researching a new medical diagnosis, planning a career change, or conducting a deep literature review, they often lack the precise keywords needed for a single query. In these scenarios, the search process itself becomes a cognitive journey of sense-making.
The Cognitive Gap in Information Retrieval
The fundamental problem addressed in this work is the gap between simple data retrieval and the cognitive process of synthesis. Most search engines are designed for a "turn-taking" model: the user provides a query, and the system returns a list. However, this model does not account for users who need to explore unfamiliar information landscapes or develop new mental models.
By synthesizing decades of research from information science, human-computer interaction (HCI), and psychology, the authors build a formal framework for Exploratory Search Systems (ESS). They categorize search behavior into three primary modes:
- Lookup: Fact-finding and verification.
- Learning: Comparison, mapping, and object identification.
- Investigation: Analysis, synthesis, and evaluation.
Theoretical Frameworks and System Design
To support these complex behaviors, the authors integrate several key theories of information seeking:
- Berrypicking: A model where the user’s information need changes as they encounter new information, picking up "bits" of data along a path rather than finding a single "best" document.
- Information Foraging: An analogy to biological foraging, where users navigate based on "information scent"—cues that suggest a particular path will lead to valuable content.
Features of Exploratory Search Systems (ESS)
An ESS is not merely a faster search engine; it is a system designed to augment human intelligence. The authors propose a taxonomy of technical features required for such systems:
- Dynamic Facets and Metadata Filtering: Allowing users to slice and dice information in real-time to see how different concepts relate.
- Visualizations: Moving beyond list-views to support insight and pattern recognition.
- Multi-tactical Interaction: Supporting "opportunistic" shifts in strategy, such as pivoting from a keyword search to browsing a related category without losing context.
- Contextual Awareness: Treating user history and task stage as first-class citizens in the search interface.
Redefining Success and Evaluation
One of the most significant contributions of this work is the call to shift how we evaluate search systems. In traditional Information Retrieval (IR), success is measured by precision, recall, and "time to success" (how quickly a user leaves the system).
In exploratory search, these metrics can be misleading. A user who spends more time in a system but emerges with a deeper understanding of a topic represents a success, not a failure. Therefore, the authors propose new evaluation frameworks that focus on user learning, task progression, and the quality of the knowledge synthesized during the session.
Limitations and Future Directions
Despite the progress in defining the field, several challenges remain for the widespread adoption of ESS:
- The Evaluation Gap: Measuring "insight" or "learning" is significantly more complex and resource-intensive than measuring clicks.
- Task Adaptation: Current systems are largely static. Future systems must be able to automatically detect when a user has shifted from a simple lookup to an exploratory mode and adapt the user interface accordingly.
- Collaborative Search: While search is often seen as a solitary act, the authors point toward a future where social and collaborative tools are integrated directly into the exploration process.
- Ubiquity: Transitioning these complex capabilities from the desktop to mobile and ubiquitous computing environments remains a major design hurdle.
Conclusion
Exploratory search represents a shift from seeing search as a retrieval task to seeing it as a cognitive process. By fostering a symbiotic relationship between the human and the machine, Exploratory Search Systems can better support the high-level synthesis and investigation required in the modern information age.
Summary updated