Comments on: Big Huge Thesaurus API

By: John

John — Thu, 03 Apr 2008 00:59:07 +0000

For those of you following along, the API now:

- searches harder to find a match by dropping common endings (ing, ed, s) when it can’t find an exact match.

- has better error codes in the form of HTTP response codes

- has a version 2 of the API with part of speech data

- sorts results by “most common” words first

And you can now purchase upgraded service directly from the site. See the API key request page for pricing.

By: Rob

Rob — Wed, 26 Mar 2008 15:44:12 +0000

Very cool. It would be useful to me (read: willing to pay for access) if it did the following:

1) Separated by type (noun, verb, adverb, etc.)

2) Separated by tense (past, future, -ing, etc.)

3) Prioritized by closest match to original keyword

In other words, if I enter the word ‘recognize’, it returns a list like:
accept
accost
accredit
acknowledge
address
agnise
agnize
appreciate
be
call back
call up
certify
cognise
…

It would be great if the ones at the top of the list were things like:
acknowledge
accredit
appreciate

Not sure how possible that would be, but I have my wallet out!

By: John

John — Mon, 10 Mar 2008 14:05:18 +0000

I’ve got that data. The reason it’s like this is that I was starting with the plain text output and didn’t want to complicate it and I wanted all the various output formats to match.

For the XML output, I could add p=”noun” as an attribute for each word element. But for the plain text output? word,partofspeech? I don’t know. Then again, maybe the different formats don’t have to match…

By: Avinash

Avinash — Mon, 10 Mar 2008 04:53:33 +0000

I would like to make a suggestion: it would be really cool for the XML version to group the results into nouns, verbs etc and returns that to the user.

eg.

data1
data2

data3
data4

By: John

John — Fri, 29 Feb 2008 18:39:16 +0000

Good points. Thanks for letting me know your thoughts. I’ll probably add some of your suggestions especially re: tenses. I’m not sure about the ings though. Ideally each of those unique versions would be added to the dictionary. That is, hosting should be in the dictionary rather than just returning the listing for host.

By: dmitry sherman

dmitry sherman — Fri, 29 Feb 2008 17:14:42 +0000

Hey John, I’ve just tested the API or actually the dictionary itself; it has some lacks…
it doesn’t strip the present progressive suffix ‘ing’ for some of the words(for instance ‘hosting’) and it doesn’t strip the past tense suffix ‘ed’(i.e ‘played’) (it’s only what i checked). Sometimes it provides results and I guess it happends because of manually added word forms. (’working’).
For comparison the Oxford version in Babylon which which do the job.

I thinkabout some client side processing for stripping, but what about the words that have special meaning in tense inflection ? Or what about the words that were already added?