edu.stanford.nlp.sentiment

Class ReadSentimentDataset

java.lang.Object

edu.stanford.nlp.sentiment.ReadSentimentDataset

public class ReadSentimentDataset
extends java.lang.Object

Reads the sentiment dataset and writes it to the appropriate files.

Author:

John Bauer

Method Summary

Modifier and Type	Method and Description
static Tree	convertTree(java.util.List<java.lang.Integer> parentPointers, java.util.List<java.lang.String> sentence, java.util.Map<java.util.List<java.lang.String>,java.lang.Integer> phraseIds, java.util.Map<java.lang.Integer,java.lang.Double> sentimentScores, PTBEscapingProcessor escaper, int numClasses)
static void	main(java.lang.String[] args) This program converts the format of the Sentiment data set prepared by Richard, Jean, etc.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

convertTree

public static Tree convertTree(java.util.List<java.lang.Integer> parentPointers,
                               java.util.List<java.lang.String> sentence,
                               java.util.Map<java.util.List<java.lang.String>,java.lang.Integer> phraseIds,
                               java.util.Map<java.lang.Integer,java.lang.Double> sentimentScores,
                               PTBEscapingProcessor escaper,
                               int numClasses)

main

public static void main(java.lang.String[] args)

This program converts the format of the Sentiment data set prepared by Richard, Jean, etc. into trees readable with the normal TreeReaders.
An example command line is
java edu.stanford.nlp.sentiment.ReadSentimentDataset -dictionary stanfordSentimentTreebank/dictionary.txt -sentiment stanfordSentimentTreebank/sentiment_labels.txt -tokens stanfordSentimentTreebank/SOStr.txt -parse stanfordSentimentTreebank/STree.txt -split stanfordSentimentTreebank/datasetSplit.txt -train train.txt -dev dev.txt -test test.txt
The arguments are as follows:
-dictionary, -sentiment, -tokens, -parse, -split Path to the corresponding files from the dataset
-train, -dev, -test Paths for saving the corresponding output files
Each of these arguments is required.
Macro arguments exist in -inputDir and -outputDir, so you can for example run
java edu.stanford.nlp.sentiment.ReadSentimentDataset -inputDir /u/nlp/data/sentiment/stanfordSentimentTreebank -outputDir . or if you're on my laptop: java edu.stanford.nlp.sentiment.ReadSentimentDataset -inputDir ../data/sentiment/stanfordSentimentTreebank -outputDir . or if you're on my desktop (please go away): java edu.stanford.nlp.sentiment.ReadSentimentDataset -inputDir ../../extern_data/sentiment/stanfordSentimentTreebank -outputDir ../../extern_data/sentiment/sst/fiveclass java edu.stanford.nlp.sentiment.ReadSentimentDataset -inputDir ../../extern_data/sentiment/stanfordSentimentTreebank -outputDir ../../extern_data/sentiment/sst/binary -numClasses 2