Search Engine with 3D Visualization of Results

Abstract

In this thesis I have designed a search engine prototype to study 3D visualization of search results. The prototype is made up of a crawler, a search engine server, and a client for visualizing the search results. The search engine is designed to use the crawler for automatic downloading, indexing, and retrieval on document collections retrieved from the World Wide Web.For document retrieval the search engine incorporates both the vector model and the latent semantic indexing model.The client application is based on SimVis, the 3D concept map visualization program developed by Ole Kristian Hvaale.There are two hypothesis presented in this thesis:1.A search engine utilizing 3D visualization is an effective way to automatically organize documents. 2.3D visualization is superior to linear list representation since relationships between documents can be seen.Both have been verified by tests performed on small document collections. The prototype implemented have it's limitations when it comes to collections which are larger than approximately 300 documents. But it is sufficiently good as proof of concept and for small scale usage.A live demo of the system can be found at http://cubix.idi.ntnu.no/simvis/SimVis.html and http://ldata.bjorli.com/simvis/SimVis.html