It's All Semantic With the New Text-Processing API

Now I don’t have a master’s degree in Natural language processing, and you just might need one to get your hands dirty with this API.  I see the API as offering a mid-level utility for incorporation in a web app.  You might take text samples from your source, feed them through the Text-Processing API and analyze those results a bit further before presenting anything to your user.

This offering appears to be the result of a one man effort. Jacob Perkins designed his API as RESTful with JSON responses.  It’s free and open for the meantime, but it sounds like Perkins may polish the service a bit and start charging for access.  There could be a real market here since only a handful of the 58 semantic APIs in our directory offer results at the technical level.

There are four main features on display here.  The first is sentiment analysis, which will give you a negative, positive or neutral rating on the feeling of the text sample.  Stemming is the second feature and it is seemingly a first step in processing a block of text during which the root, or stem, of a word is identified.  My guess is that stemming normalizes the input a bit, narrowing down the possibilities that your application would need to consider.  Also quite technical, features three and four--“part of speech tagging and chunking” and “phrase extraction and named entity recognition”--again give you results for use in a next step of language processing.

Be sure to read the next API article: uShip Gets It Moving With Bidding


Comments (5)


Thanks for the write-up. I'd also like to note that the tagging API is currently being used by Reworder, in conjunction with WordNik:

[...] API got a write up (and a nice traffic boost) from Garrett Wilkin (@garrettwilkin) on Share this: Tagged as: erlang, nlp, nltk, python Leave a comment blog comments powered [...]

[...] It offers the familiar features of previous semantic analysis APIs such as those provided by and Syllabs.  The Saplo API is aimed toward articles.  Each article can be tagged, categorized [...]

[...] tuned to find out.  The service kind of reminds me of, which we profiled back in April 2011. Related ProgrammableWeb ResourcesLearn more Roistr API [...]