gkotsis / negation-detection

Negation detection NLP tool. If you use the code, please cite George Gkotsis, Sumithra Velupillai, Anika Oellrich, Harry Dean, Maria Liakata and Rina Dutta. Don't Let Notes Be Misunderstood: A Negation Detection Method for Assessing Risk of Suicide in Mental Health Records, Computational Linguistics and Clinical Psychology 2016

Negation Resolution

An NLP tool that performs negation resolution on a sentence level. It takes as input a sentence and a target-keyword. It returns True for affirmed keywords, False for negated and None for keywords not found. The tool makes use of Stanford's CoreNLP constituency trees.

Installation

1. git clone https://github.com/gkotsis/negation-detection

2. Download and extract CoreNLP.

3. Install stanford_corenlp_pywrapper

4. Install through requirements.txt:

   cd negation-detection
   pip install -r requirements.txt
   

5. Edit settings.json. Make sure you keep the leading slashes in the directory names.

##Example

import negation_detection
sentence = "ZZZZ reported no recent periods of low mood, discussed how in the past she made many suicide attempts"
negation_detection.predict(sentence, 'suicide')

Reference

George Gkotsis, Sumithra Velupillai, Anika Oellrich, Harry Dean, Maria Liakata and Rina Dutta. Don't Let Notes Be Misunderstood: A Negation Detection Method for Assessing Risk of Suicide in Mental Health Records, Workshop on Computational Linguistics and Clinical Psychology, NAACL 2016.

paper

supplement