Visual Exploratory Search of Relationship Graphs on Smartphones

Jianquan Ouyang, Hao Zheng, Fanbin Kong, Tianming LiuPLoS 2013 · DOI

Abstract

This paper introduces the Visual Exploratory Search of Relationship Graphs on Smartphones (VESRGS) framework. VESRGS moves beyond traditional "list of links" search paradigms by automatically constructing semantic relationship graphs from web meta-search results. By combining cloud-based graph inference with a multi-touch mobile interface, the system enables users to discover non-obvious connections between concepts—such as links between specific foods and health conditions—in a personalized, exploratory environment.

Traditional search engines are optimized for "lookup" tasks where a user has a specific question and receives a ranked list of documents. However, this model often fails during exploratory search, where a user aims to learn, investigate, or discover relationships within a complex topic. The authors identify three primary challenges in current mobile search: information overload, hidden relationships between results, and the physical constraints of small smartphone screens that make scrolling through endless lists inefficient for knowledge discovery.

The VESRGS Framework

The VESRGS framework utilizes a client-server architecture to bridge the gap between heavy data processing and mobile usability. The core of the system is a three-step graph inference process.

The three-step process for inferring semantic relationships from web meta-search results, moving from keyword pairs to a comprehensive triple set.
  1. Graph Inference via Meta-Search: The system takes domain keywords (e.g., "Foods" and "Diseases") and queries Google, Yahoo!, and Bing for all possible pairs. It uses statistical Natural Language Processing (NLP) to calculate the probability of co-occurrence. If two terms appear together frequently across the web, a weighted edge is created between them in a semantic graph.
  2. Cloud-to-Mobile Pipeline: A cloud server handles the computationally expensive tasks of crawling and graph construction. A Result Processing System (RPS) then tailors these graphs for mobile display by grouping nodes based on density and importance.
  3. Personalized Filtering: The system models user interests as a graph. By performing a graph matching (intersection) between the global relationship graph and the user's profile, the system highlights the most relevant "Graph of Interest" (GOI), reducing visual clutter.

Knowledge Discovery and Results

The authors validated the system using real-world data from the USDA (food terms) and NIH (disease terms). In a subjective evaluation with volunteers, VESRGS received an average usability score of 4.6/5, significantly outperforming traditional search engines (averaging ~3.5) for finding relevant relationships.

A visualized relationship graph generated by VESRGS connecting food items (green nodes) to health conditions (red nodes).

The system successfully identified known but non-obvious relationships, such as the link between pepper consumption and stroke prevention, and the role of zinc in maple syrup for prostate health. A key insight from the study is the shift from "recall" to "recognition"; it is cognitively easier for a user to recognize a relevant connection on a visual map than to recall the perfect keyword to find it through a standard search.

Limitations and Future Work

While powerful, the system currently identifies co-occurrence (correlation) rather than causation. Full elucidation of these links requires expert biomedical validation. Future work aims to move toward real-time graph inference for arbitrary user-inputted keywords and expanding the framework to other complex domains, such as environmental factors or professional healthcare databases like PubMed.

Summary updated

Abstract

This paper presents a novel framework for Visual Exploratory Search of Relationship Graphs on Smartphones (VESRGS) that is composed of three major components: inference and representation of semantic relationship graphs on the Web via meta-search, visual exploratory search of relationship graphs through both querying and browsing strategies, and human-computer interactions via the multi-touch interface and mobile Internet on smartphones. In comparison with traditional lookup search methodologies, the proposed VESRGS system is characterized with the following perceived advantages. 1) It infers rich semantic relationships between the querying keywords and other related concepts from large-scale meta-search results from Google, Yahoo! and Bing search engines, and represents semantic relationships via graphs; 2) the exploratory search approach empowers users to naturally and effectively explore, adventure and discover knowledge in a rich information world of interlinked relationship graphs in a personalized fashion; 3) it effectively takes the advantages of smartphones' user-friendly interfaces and ubiquitous Internet connection and portability. Our extensive experimental results have demonstrated that the VESRGS framework can significantly improve the users' capability of seeking the most relevant relationship information to their own specific needs. We envision that the VESRGS framework can be a starting point for future exploration of novel, effective search strategies in the mobile Internet era.
BibTeX
@inproceedings{ouyang2013visual,
  author = {Jianquan Ouyang and Hao Zheng and Fanbin Kong and Tianming Liu},
  title = {{Visual Exploratory Search of Relationship Graphs on Smartphones}},
  booktitle = {PLoS},
  year = {2013},
  doi = {10.1371/journal.pone.0079379},
}
Figure