[
https://issues.apache.org/jira/browse/MAHOUT-242?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12806357#action_12806357
]
Drew Farris commented on MAHOUT-242:
bq. Hey Drew, I'm not much of a maven guy - what's the maven-foo you use to get
this running?
Jake, my mistake, I should have included updated docs when I updated the patch.
I eliminated the code to read plain text files from a directory, so you would
need to begin by producing a SequenceFileText,Text (document id, document) as
input. Robin's utility in mahout-examples,
o.a.m.text.SequenceFilesFromDirectory can do this. Run the following from the
'examples' directory;
{code}
mvn -e exec:java
-Dexec.mainClass=org.apache.mahout.text.SequenceFilesFromDirectory
-Dexec.args=--parent (...input directory..) --outputDir (..output directory..)
--charset UTF-8
{code}
Once you have the sequence file to use as input, run the following from the
'examples' directory as well.
{code}
mvn -e exec:java
-Dexec.mainClass=org.apache.mahout.nlp.collocations.llr.CollocDriver
-Dexec.args=--input (..path-to-input..) --output (..path-to-output..) -w
{code}
Once the driver class is run, the collocations will be in
(output-directory)/colloc/part-0 as plaintext. They can be sorted by LLR
score using the same sort command I included in the previous comment above (I
have a question about this below).
FWIW, I need to re-submit the patch to clean up some of the pom changes to
MAHOUT-215 were applied.
A couple follow up questions while you're looking at this:
Currently I just dump the results of the second pass to a file, not sorted by
LLR score. I could sort by LLR by sending it through another pass with an
identity mapper and a reducer but I suspect that's probably pretty inefficient.
Is there a better way to sort the output of the second pass by LLR?
If I only wanted to emit the top 1-10% of the collocs (user configurable), how
would I tell the reducer to stop emitting results at a certain point (or is yet
another pass needed to achieve something like this?)
Would it be better to emit a sequencefileLongWritable,Text instead of a text
file as the output from the final pass?
LLR Collocation Identifier
--
Key: MAHOUT-242
URL: https://issues.apache.org/jira/browse/MAHOUT-242
Project: Mahout
Issue Type: New Feature
Affects Versions: 0.3
Reporter: Drew Farris
Assignee: Jake Mannix
Priority: Minor
Attachments: MAHOUT-242.patch, MAHOUT-242.patch,
mahout-colloc.tar.gz, mahout-colloc.tar.gz
Identifies interesting Collocations in text using ngrams scored via the
LogLikelihoodRatio calculation.
As discussed in:
*
http://www.lucidimagination.com/search/document/d051123800ab6ce7/collocations_in_mahout#26634d6364c2c0d2
*
http://www.lucidimagination.com/search/document/b8d5bb0745eef6e8/n_grams_for_terms#f16fa54417697d8e
Current form is a tar of a maven project that depends on mahout. Build as
usual with 'mvn clean install', can be executed using:
{noformat}
mvn -e exec:java -Dexec.mainClass=org.apache.mahout.colloc.CollocDriver
-Dexec.args=--input src/test/resources/article --colloc target/colloc
--output target/output -w
{noformat}
Output will be placed in target/output and can be viewed nicely using:
{noformat}
sort -rn -k1 target/output/part-0
{noformat}
Includes rudimentary unit tests. Please review and comment. Needs more work
to get this into patch state and integrate with Robin's document vectorizer
work in MAHOUT-237
Some basic TODO/FIXME's include:
* use mahout math's ObjectInt map implementation when available
* make the analyzer configurable
* better input validation + negative unit tests.
* more flexible ways to generate units of analysis (n-1)grams.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
