lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Grant Ingersoll (JIRA)" <>
Subject [jira] Updated: (LUCENE-1103) WikipediaTokenizer
Date Wed, 02 Jan 2008 16:12:34 GMT


Grant Ingersoll updated LUCENE-1103:

    Fix Version/s: 2.3

Patch shortly.  This will be all new code, other than minor changes to include javadocs. 
I am going to create contrib/wikipedia, as there are probably other things that can go in
here once the seed is started.

> WikipediaTokenizer
> ------------------
>                 Key: LUCENE-1103
>                 URL:
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: Analysis
>            Reporter: Grant Ingersoll
>            Assignee: Grant Ingersoll
>            Priority: Minor
>             Fix For: 2.3
> I have extended StandardTokenizer to recognize Wikipedia syntax and mark tokens with
certain attributes.  It isn't necessarily complete, but it does a good enough job for it to
be consumed and improved by others.
> It sets the Token.type() value depending on the Wikipedia syntax (links, internal links,
bold, italics, etc.) based on my pass at
> I have only tested it with the benchmarking EnwikiDocMaker wikipedia stuff and it seems
to do a decent job.
> Caveats:  I am not sure how to best handle testing, since the content is licensed under
GNU Free Doc License, I believe I can't copy and paste a whole document into the unit test.
 I have hand coded one doc and have another one that just generally runs over the benchmark
Wikipedia download.
> One more question is where to put it.  It could go in analysis, but the tests at least
will have a dependency on Benchmark.  I am thinking of adding a new contrib/wikipedia where
this could grow to have other wikipedia things (perhaps we would move EnwikiDocMaker there????)
and reverse the dependency on Benchmark.
> I will post a patch over the next few days.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message