lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joshua O'Madadhain <>
Subject RE: compiling example code
Date Wed, 14 Nov 2001 02:28:11 GMT
On Tue, 13 Nov 2001, Alex Murzaku wrote:

> Are you using ant? By just using "ant demo" from the lucene root
> directory everything goes fine. Make sure you have the latest ant
> (1.4).

I have no idea what ant is or what it's supposed to be for; I'll do a web
search to see if I can get one (an idea, that is), but a brief explanation
would be appreciated.  However, I would hope that it shouldn't be
necessary to use a separate tool (aside from javac, that is) in order to
just build some fairly basic sample code.
> And also, I would be very interested in seeing what you are doing. At
> my previous job, we started building a clusterizer around lucene but
> turned out to be a little bit slow, so, we ended up building our own
> similarity measuring (manhattan distance between word vectors).

The basic idea is that I'm investigating a technique that combines fuzzy
term expansion and fuzzy reranking of query results: the term expansion
should help improve recall, and the reranking should help to improve
precision.  You do this by creating a weighted graph that represents word
similarities in some fashion and adding to the query those terms that are
sufficiently close to the query terms (using your favorite definitions of
"sufficient" and "close").  On the back end, you include in the ranking
of each document a factor that takes account of just how fuzzy a match the
terms in the document are for the query terms.

How fast this will be, especially for this prototype, I don't know; right
now I'm more interested in its effect on recall and precision than
anything else.  It does require some additional overhead up front (when
the index is created) but shouldn't require too much extra work on a
query-by-query basis.  We shall see.  :)

Joshua Per
    Joshua Madden: Information Scientist, Musician, Philosopher-At-Tall
 It's that moment of dawning comprehension that I live for--Bill Watterson
My opinions are too rational and insightful to be those of any organization.

> Two of these files compile without error.  However, when I attempt to
> compile, I get the following error:
> C:\lucene-1.2-rc2\src\demo\org\apache\lucene\ cannot
> resolve symbol
> symbol  : variable FileDocument
> location: class org.apache.lucene.IndexFiles
>       writer.addDocument(FileDocument.Document(file));
>                          ^
> 1 error
> I don't understand why I'm getting this error; FileDocument has already
> been compiled, and the class is in the same directory as the other
> documents.  Perhaps I'm just missing something basic about using packages,
> but it all seems plausible.
> Has anyone managed to get these examples to build, and if so, how?

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

View raw message