This repository is meant to access APES summarization evaluation metric on the data published in TAC AESOP 2011.
In order to run the preprocessing script on the data please run the following command:
python apes_on_tac2011.py --mode preprocess
Three examples from the original AESOP dataset are available in the examples
folder.
In order to obtain relevant AESOP 2011 data please refer to AESOP 2011 website and submit a request. The data can not be added to this repository.
To run the QA you need a QA stream that expects queries.pkl
file and writes answering accuracy in rewards.txt
.
You can find a trained QA here, so it is required you run the QA stream before running the answering_questions process.
python apes_on_tac2011.py --mode answer_questions
This will expect a QA system that reads questions from ./queries.pkl
file and writes answering accuracy in ./rewards.txt
.