Enhancing Search and Browse Using Automated Clustering of Subject Metadata
Enhancing Search and Browse Using Automated Clustering of Subject Metadata
From the abstract:
The Web puzzle of online information resources often hinders end-users from effective and efficient access to these resources. Clustering resources into appropriate subject-based groupings may help alleviate these difficulties, but will it work with heterogeneous material? The University of Michigan and the University of California Irvine joined forces to test automatically enhancing metadata records using the Topic Modeling algorithm on the varied OAIster corpus.
Clustering can not only serve to help the searcher find what they are (or think they are) looking for but we think it’s just as valuable to present ideas, concepts, names, trends, etc. that the searcher might not see by browsing one record at a time.
Notes 1: We have posted about OAIster many times on ResourceShelf. This database aggregates over 12 million records from more than 854 contributors. A very useful and important resource.
Notes 2: Of course no conversation about clustering can take place without mention Raul Valdes-Perez, Co-Founder and CEO of Vivisimo. Vivisimo powers Clusty and offers dynamic clustering for enterprise search. While white papers can only say so much, this paper from Vivisimo (PDF) about information overload and what they call “selective ignorance” is worthy of your attention. You can learn more by browsing the Vivisimo site, reading the company blog (with commentary from Valdes-Perez) and using some of their projects beyond Clusty.
Examples:
+ Clustermed.info
Clustering (and note the many ways to cluster) PubMed since this search is using controlled fields.
+ The USA government portal search at USASearch.gov/
+ New Zealand Government Search
+++ and some often missed Clusty features:
+ Cluster Jobs (via Indeed.com)
+ Cluster Blog Posts
and the Cool Clusty Clouds from the Clusty Labs.
Notes 3: Since Ask.com provides “Zoom Related Info” that offers terms to narrow and/or expand a search (based on concepts) and in some cases, show related names (again, based on concepts), I’ll keep my comments to this example and some of what has been written elsewhere. 1 ||| 2 ||| 3.
Btw, Zoom Related Search is also available with Ask Image search.
Here’s an example using a search for Paul McCartney). Note the Zoom Related Results in the left pane.
Thanks to Pete Weiss for the news tip.
This entry was posted
on Saturday, August 11th, 2007 at 1:07 am and is filed under Info Management and Retrieval, Information Industry, Information Seeking, Search News.
You can follow any responses to this entry through the RSS 2.0 feed.
Responses are currently closed, but you can trackback from your own site.