- Download pSenti lexicon based utility
- Python 2.7
- Keras >= 2.0.5
- CNTK 2.2
- Unit Test Data
python DiscoverLexicon.py -d amazon_kitchen # to extract full kitchen domain lexicon
python DiscoverLexicon.py -d amazon_video # to extract full kitchen domain lexicon
python DiscoverLexicon.py -d amazon_electronics # to extract full kitchen domain lexicon
Wikiled.Sentiment.ConsoleApp.exe test -Articles="[Path]\Kitchen_exp.xml" -Out=.\kitchen_psenti
Wikiled.Sentiment.ConsoleApp.exe test -Articles="[Path]\Kitchen_exp.xml" -Out=.\kitchen_own -Weights="[Path]\words_Amazon_Kitchen.csv" -FullWeightReset
Wikiled.Sentiment.ConsoleApp.exe test -Articles="[Path]\Kitchen_exp.xml" -Out=.\kitchen_t_el -Weights="[Path]\words_Amazon_Electronics.csv" -FullWeightReset
Wikiled.Sentiment.ConsoleApp.exe test -Articles="[Path]\Kitchen_exp.xml" -Out=.\kitchen_t_vi -Weights="[Path]\words_Amazon_Video.csv" -FullWeightReset
Can be downloaded from (http://ai.stanford.edu/~amaas/data/sentiment/)
python DiscoverLexicon.py -d imdb -c 0.7 -b # Induce Sentiment Lexicon
Wikiled.Sentiment.ConsoleApp.exe bootimdb -Words="[path]\words_imdb.csv" -Path="[path]" -Destination="[path]" -BalancedTop=0.8 # Bootstrap Training dataset
Python Sentiment.py -d imdb -a lstm -n 2 -p # Train sentiment classifier
python DiscoverLexicon.py -d amazon -c 0.7 -b
Wikiled.Sentiment.ConsoleApp.exe boot -Words="[path]\words_amazon.csv" -Path="[path]\unlabel.txt" -Destination="[path]" -BalancedTop=0.8
Python Sentiment.py -d imdb -a lstm -n 2 -p
In brackets listed options for binary and multiclass classification. If necesary you can bootstrap neutral class with pSenti
python DiscoverLexicon.py -d semeval -c 0.7 -b # Induce Sentiment Lexicon
Wikiled.Sentiment.ConsoleApp.exe semboot -Words="[path]\words_semeval.csv" -Path="[path]" -Destination="[path]\train.txt" -InvertOff [-Neutral] # Bootstrap Training dataset
Python Sentiment.py -d semeval -a lstm -n [2 or 5] -p # Train sentiment classifier