AlchemyAPI Semantic Analysis Plugin

[shaunroach]

This is a thread for the AlchemyAPI plugin to Gephi where anyone can give comments or feature requests.

You can download the plugin here: http://gephi.org/plugins/alchemyapi-plugin/

Today we submitted a new version that will create edge connections between text sources that have common entities, so you can run the plugin over an archive of text and see what shared concepts and interconnections emerge.

Here's the full description of the plugin:
This plugin will generate a tree of entities(people,places,things) extracted from a url, text, or html file.
Using natural language processing technology from AlchemyAPI, the following entities will be detected in the source document and displayed:
Anniversary, City, Company, Continent, Country, EntertainmentAward, Facility, FieldTerminology, FinancialMarketIndex, GeographicFeature, HealthCondition, Holiday, Movie, MusicGroup, NaturalDisaster, Organization, Person, PrintMedia, RadioProgram, RadioStation, Region, Sport, StateOrCounty, Technology, TelevisionShow, TelevisionStation

Shaun

[pbittner]

This is a very interesting plugin, thank you for sharing it. I tried it out on a couple of web pages and it gives very nice results. It would be even more interesting if a weight was assigned to each node (based on the frequency of the word in the text for example) so that we could rank the nodes using that parameter and identify the main entities. Would that be possible?

[shaunroach]

Yes, good suggestion. The next version of the plugin will have a relevance attribute and a count attribute that you can use to size/color/filter the nodes. I think that will provide the ranking you suggest.

I'll post here once it gets released, it should be out in the next day or two. Thanks for trying it out.

Shaun

[shaunroach]

Blog post on the Gephi blog about the semantics plugin: http://gephi.org/2010/semantic-plugin-alchemyapi/

Highlights a bunch of new features including the ability to make a graph from text archives.