kbnlp.hs

Mark Watson's Haskell NLP Experiments

Copyright 2014 by Mark Watson. All rights reserved.

License: AGPL version 3 license. (Note: I own the copyright of all of the code in this project; contact me if you need a commercial license.)

I am currently working on using, when possible, DBPedia URIs as identifiers for entities detected in text. As is much of this project, this is a work in progress.

Note: this project is no longer supported - see newer version in code for my Haskell book

Please use the code in the NlpTool directory at https://github.com/mark-watson/haskell_tutorial_cookbook_examples

Credits

I use the Haskell stemmer written by Dmitry Antonyuk and the sentence splitting code written by Eric Kow.

Thanks!!

Generated code

There are a fair number of Haskell "source" files that were generated by Ruby scripts that are not included in this project.

These files (e.g., CityNamesDbpedia.hs, Category1Gram.hs, Category2Gram.hs, etc.) mostly use Data.Map.fromList to create in-memory maps for lingusitic and other data. These files are not particularly interesting. The more interesting code is found in the top level files Summarize.hs, Entities.hs, etc.

Using stack

stack build stack exec nlpdemo

then open http://localhost:3000