Find us on Google+

Mar 13, 2009

Eyeplorer : a new visual search engine

Eyeplorer Overview in English


Eyeplorer is a new visual search engine created by a Germany based company. It lets you explore visually Wikipedia contents in a highly intuitive way. After entering the subject of your search in the center of the map you can see the links and connections between articles. You can also drag & drop the chosen facts to the notepad on the right side and copy the facts to clipboard or share with a friend via email.

4 comments:

chris said...

The technology looks very poor and crappy. Its based on simple pattern matching and collocation analysis. The technology behind is also based on a simple activation spreading network which produces the k-graphs - the term here is k-matrix. The graphs are static and precalculated. We must calculate the whole graph new if something changes. Also the content base must contain valid text. The system is not be able to search in webpages, cause the lingustic system behind must analyse sentences. Webpages has quite more problematic text patterns. Sentences are not simple to recognize and extract, it is very domain dependend. Also there is not idea yet to prevent spam and SEO stuff.The system also has yet no ability to recognize invalid statements (Subject-Predicate-Object) like “A headset is human”. In opposite to the suggestion of the name (vionto == visual ontology) the system does not based on any ontology which follows an official definition.

Philippe Boukobza said...

Thank for sharing your expertise. Do you know any visual application that is able to cope with sentences, to recognize invalid statements or to prevent SEO stuff ?

Georg Rehm said...

First of all, thanks for mentioning eyePlorer.com! A few words to the other comment: what "chris" is speculating about with regard to our technical processing is not true at all. I'm sure you'll understand that I'm unable to go into details but, among other components, we develop and use quite sophisticated natural language processing modules and several methods from AI research.

Philippe Boukobza said...

Thanks Georg for your comment. It's really great that people like you enter this blog. I stay impressed by Eyeplorer and the innovation of this project. Eyeplorer belongs for me to the kind of tools we will use daily in the Semantic Web.