Get highest frequency terms from Lucene index

Question

i need to extract terms with highest frequencies from several lucene indexes, to use them for some semantic analysis.

So, I want to get maybe top 30 most occuring terms(still did not decide on threshold, i will analyze results) and their per-index counts. I am aware that I might lose some precision because of potentionally dropped duplicates, but for now, lets say i am ok with that.

So for the proposed solutions, (needless to say maybe) speed is not important, since I would do static analysis, I would put accent on simplicity of implementation because im not so skilled with Lucene and cant wrap my mind around some concepts of it..

I can not find any code samples from something similar, so all concrete advices (code, pseudocode, links to code samples...) Appreciate all the advices!

Thank you!

Solution 1

Have a look at this: http://sujitpal.blogspot.com/2009/02/summarization-with-lucene.html

The class in this page hascomputeTopTermQuery method which you should be easily able to retrofit for going over multiple indexes.

Solution 2

A very simple way would be to use Luke. On the 'Overview' tab, there is a 'Show top terms' button that can be used for what you need.

Comments

[Julia]

Thanx! Exactly what I needed!

[Thangnv]

Hi mindas! I using lucene 4.4 so it don't have terms() method ? help me, please!

[mindas]

@Thangnv you might want to open a separate thread for this, or email Lucene Java mailing list. I don't have Lucene 4.4 running on hand. And the time has been on short supply these days ;-(