Skip to content

Latest commit

 

History

History
83 lines (46 loc) · 3.21 KB

README.md

File metadata and controls

83 lines (46 loc) · 3.21 KB

Maui-2.0

Maui 2.0 is a keyword extraction tool, based on Maui, a topic indexing tool.

##Improvement

Major improvement over original Maui:

  • Upgraded one of its dependent libraries, Wikipedia Miner from 1.1 to 1.2, resulting in much faster program execution speed(almost 10 times faster).
  • Augmented with keyword extraction Java API. Original one has only commandline tool.

##Usage

###Preparing Wikipedia Miner Maui 2.0 depends Wikipeida Miner 1.2, which requires the user to build the Wikipedia summary database.

Instruction on how to set up Wikipedia Miner 1.2 is located at wikipedia-miner-setup.

If you need to Wikipedia Dump summary file of July, 2014, please contact me(see the address at the very bottom).

###Build pacakge > cd ant > ant package

###Buidling the model

Run

> java -cp "lib/*:build/jar/maui-2.jar" maui.main.MauiModelBuilder -l {path-to-trainig-data} -m {path-to-model-file} -v wikipedia -c {path-to-wikipedia-miner-configuration-file}

  • For wikipedia miner configuration file, refer to here
  • A typical trianing data path would be data/wikipedia_indexing/train/

An example:

> java -cp "lib/*:build/jar/maui-2.jar" maui.main.MauiModelBuilder -l data/wikipedia_indexing/train -m test -v wikipedia -c {path-to-wikipedia-miner-configuration-file}

###Keyword extraction(Command line)

Given a list of text files for which you would the keywords to be extracted, an example usage is:

> java -cp "lib/*:src" maui.main.MauiTopicExtractor -l {path-to-test-data} -m {path-to-model-file} -v wikipedia -c {path-to-wikipedia-miner-configuration-file}

  • test data path is the directory containing text files each being a document with its keywords to be extracted.
  • Note text files in test data path should have extension .txt

An example:

> java -cp "lib/*:build/jar/maui-2.jar" maui.main.MauiTopicExtractor -l data/wikipedia_indexing/test -m test -v wikipedia -c {path-to-wikipedia-miner-configuration-file}

###Keyword extraction(Java API)

The function can also be called through Java API. An Scala example is:

import maui.main.MauiTopicExtractor

val opts = Array("-v", "wikipedia", "-m", "/path/to/model-file", "-c", "/path/to/wikipedia-miner/config.xml")

val extractor = new MauiTopicExtractor()
val text = "The Liber Eliensis (\"Book of Ely\") is a 12th-century English chronicle and history, written in Latin. Composed in three books, it was written at Ely Abbey on the island of Ely in the fenlands of eastern Cambridgeshire. Ely Abbey became the cathedral of a newly formed bishopric in 1109."

extractor.setOptions(opts)
extractor.loadModel()

extractor.configMauiFilter()

println("Keyphrases are: " + extractor.extractKeyphrasesFromText(text).toList.toString)

To run the example, just execute:

scala -classpath "lib/*:build/jar/maui-2.jar" example.scala

##Contact

Han Xiao