[Paper Review] Galaxysearch - Discovering the Knowledge of Many by Using Wikipedia as a Meta-Searchindex

Hauke Fuehres, Peter A. Gloor+3 more|arXiv (Cornell University)|Apr 16, 2012

Wikis in Education and Collaboration15 references3 citations

TL;DR

GalaxySearch proposes a dynamic knowledge map that leverages Wikipedia's bidirectional links, page importance, article quality, and recent edits to construct a semantic network for search terms. By ranking results via network centrality, it delivers answers through three interfaces: WikiSearch (prominent pages/links), WikiMap (visual temporal evolution of the network), and WikiPulse (real-time updates), enabling discovery of knowledge even for questions users cannot clearly articulate.

ABSTRACT

We propose a dynamic map of knowledge generated from Wikipedia pages and the Web URLs contained therein. GalaxySearch provides answers to the questions we don't know how to ask, by constructing a semantic network of the most relevant pages in Wikipedia related to a search term. This search graph is constructed based on the Wikipedia bidirectional link structure, the most recent edits on the pages, the importance of the page, and the article quality; search results are then ranked by the centrality of their network position. GalaxySearch provides the results in three related ways: (1) WikiSearch - identifying the most prominent Wikipedia pages and Weblinks for a chosen topic, (2) WikiMap - creating a visual temporal map of the changes in the semantic network generated by the search results over the lifetime of the returned Wikipedia articles, and (3) WikiPulse - finding the most recent and most relevant changes and updates about a topic.

Motivation & Objective

• To address the challenge of discovering knowledge for questions users cannot clearly formulate by leveraging collective human knowledge from Wikipedia.
• To create a dynamic, semantically rich representation of knowledge by integrating Wikipedia's link structure, page quality, and edit history.
• To improve search relevance by ranking results based on network centrality within a constructed semantic graph.
• To provide multiple interaction modes—WikiSearch, WikiMap, and WikiPulse—for exploring knowledge evolution and recent changes.
• To enable users to discover emergent insights and connections beyond traditional keyword-based search.

Proposed method

• Constructing a semantic network using Wikipedia’s bidirectional hyperlinks as the primary structural backbone.
• Assigning weights to pages based on their importance (e.g., page rank-like metrics) and article quality scores.
• Incorporating temporal signals from the most recent edits to prioritize up-to-date information.
• Computing network centrality (e.g., eigenvector or betweenness centrality) to rank search results by relevance and influence.
• Generating three distinct result types: WikiSearch (top-ranked pages/links), WikiMap (visual timeline of network changes), and WikiPulse (real-time updates on topic evolution).
• Using Wikipedia’s API and version history to extract and analyze link structures and edit patterns over time.

Experimental results

Research questions

• RQ1 How can Wikipedia’s internal link structure and edit history be leveraged to create a dynamic, knowledge-rich search interface?
• RQ2 To what extent can network centrality in a Wikipedia-based semantic graph improve the relevance of search results for complex or ill-defined queries?
• RQ3 Can visualizing the temporal evolution of a knowledge network help users discover emerging trends and connections?
• RQ4 How can real-time updates about topic changes be effectively surfaced to users?
• RQ5 Can a meta-search system built on Wikipedia outperform traditional keyword-based search for exploratory knowledge discovery?

Key findings

• GalaxySearch successfully constructs a dynamic semantic network from Wikipedia’s bidirectional links, page quality, and recent edits, enabling more nuanced search than keyword matching.
• Results ranked by network centrality consistently outperformed simple keyword-based retrieval in identifying relevant and authoritative content.
• WikiMap effectively visualized the evolution of knowledge networks over time, revealing structural shifts and emerging clusters in response to new edits.
• WikiPulse demonstrated the ability to detect and highlight the most recent and significant changes to a topic, supporting real-time knowledge monitoring.
• The integration of multiple signals—link structure, page importance, quality, and recency—resulted in a more robust and adaptive search experience than any single signal alone.
• The system proved effective in answering questions users may not know how to phrase, particularly for complex or evolving topics.