Mayer123 / Multiparty-Dialog-RC

Code for the paper "Challenging Reading Comprehension on Daily Conversation: Passage Completion on Multiparty Dialog" (NAACL 2018)

Multiparty-Dialog-RC

This repository contains the code for the paper "Challenge Reading Comprehension on Daily Conversations: Passage Completion on Multiparty Dialog". See full paper here

Requirements

• Python 2.7
• Numpy >= 1.13.3
• Tensorflow >= 1.4.0
• Keras >= 2.0.9

Datasets

Our datasets with experimental splits can be found at dialog_rc_data in json format.

The original TV show transcripts in json format can be found at Character Mining project.

Word embeddings: We used Glove vectors with 100 dimentions.

Usage

    python exp.py --train_file ../dialog_rc_data/json/Trn.json 
                   --dev_file ../dialog_rc_data/json/Dev.json 
                   --embedding_file glove.6B.100d.txt
                   --model cnn_lstm_UA_DA --logging_to_file log.txt
                   --save_model model.h5 --stopwords stopwords.txt

Options

• hidden_size: default is 32.
• batch_size: default is 32.
• utterance_filters: default is 50.
• query_filters: default is 50.
• nb_epoch: default is 100.
• dropout: default is 0.2.
• learning_rate: default is 0.001.

Cite

@inproceedings{ma-etal-2018-challenging,
    title = "Challenging Reading Comprehension on Daily Conversation: Passage Completion on Multiparty Dialog",
    author = "Ma, Kaixin  and
      Jurczyk, Tomasz  and
      Choi, Jinho D.",
    booktitle = "Proceedings of the 2018 Conference of the North {A}merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers)",
    month = jun,
    year = "2018",
    address = "New Orleans, Louisiana",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/N18-1185",
    doi = "10.18653/v1/N18-1185",
    pages = "2039--2048",
}