sauravcsvt/kdd_linelisting

This repository contains the source codes and data for KDD line listing paper

KDD_linelisting

This repository contains the source codes and data for the paper entitled GELL: Automatic Extraction of Epidemiological Line Lists from Open Sources which recently got accepted in ACM SIGKDD 2017. The code for the proposed model GELL is located at linelist_code and the code for the baseline model Baseline is located at baseline_code. The codes can be executed as follows.

Code: linelist_code usage: Automated line listing [-h] -i MERSBULLETINS -v WHOVEC -ind NUMIND -o OUTPUTLL

optional arguments: -h, --help show this help message and exit -i MERSBULLETINS, --MERSbulletins MERSBULLETINS Input file containing the WHO MERS bulletins from which line list will be extracted -v WHOVEC, --whovec WHOVEC word vectors corresponding to the WHO corpus -ind NUMIND, --numind NUMIND Number of predictors to be used for extracting line each list feature -o OUTPUTLL, --outputll OUTPUTLL File where the automatically extracted line list will be dumped

Example: python ./code/ll_code.py -i ./data/WHO_KSA_MERS_bulletins.json -v ./data/WHO_vectors/WHO_SGHS_vectors.word2vec -ind 7 -o ./data/automated_ll/automated_ll_KSA_SGHS.json

Code: baseline_code usage: Baseline line listing [-h] -i MERSBULLETINS -o OUTPUTLL

optional arguments: -h, --help show this help message and exit -i MERSBULLETINS, --MERSbulletins MERSBULLETINS Input file containing the WHO MERS bulletins from which line list will be extracted -o OUTPUTLL, --outputll OUTPUTLL File where the automatically extracted line list will be dumped

Example: python ./code/ll_baseline.py -i ./data/WHO_KSA_MERS_bulletins.json -o ./data/automated_ll/automated_ll_baseline.json

Confusion matrices

The confusion matrices corresponding to the performance of each model for a clinical feature can be found at confusion_matrix.