Dialogue Ontology Relation Extraction via Constrained Chain-of-Thought Decoding
This is the code for the SIGDIAL 2024 paper Dialogue Ontology Relation Extraction via Constrained Chain-of-Thought Decoding
Data
We use the Multi-WOZ 2.1 Data-set and the Schema-Guided Dialogue data-set for which the preprocessed datasets we use can be found in the data folder.
Requirements
Install requirements using:
python3 -m pip install -r requirements.txt
Training
Train a model with a config_name given in finetuning/training/finetuning_config_list.py using:
cd finetuning
python -m training.autoregressive_finetuning --config_name ${config_name}
Inference
Do inference with a config_name given in LLM_experiments/configs.py using:
cd experiments
python -u TOD_ontology_inference.py --config_name ${config_name}
Evaluation
Evaluate with a config_name given in LLM_experiments/configs.py using:
cd experiments
python -u TOD_ontology_evaluation.py --config_name ${config_name}
For combining the predictions of different approaches, e.g. of the different one relation only prompts, use a list of config names as input:
cd LLM_experiments
python -u TOD_ontology_evaluation.py --config_name_list ${config_names[@]}
For CoT-decoding set the aggregation strategy, number of branches used and the threshold for non highest disparity branch strategies using the following flags:
--cot_aggregation_strategy ${aggregation_strategy}
--cot_disparity_threshold ${disparity_threshold}
--k_branches_to_use ${k_branches_to_use}
--cot_disparity_threshold ${disparity_threshold}
Citation
@inproceedings{vukovic-etal-2024-dialogue,
title = "Dialogue Ontology Relation Extraction via Constrained Chain-of-Thought Decoding",
author = "Vukovic, Renato and
Arps, David and
van Niekerk, Carel and
Ruppik, Benjamin Matthias and
Lin, Hsien-chin and
Heck, Michael and
Gasic, Milica",
editor = "Kawahara, Tatsuya and
Demberg, Vera and
Ultes, Stefan and
Inoue, Koji and
Mehri, Shikib and
Howcroft, David and
Komatani, Kazunori",
booktitle = "Proceedings of the 25th Annual Meeting of the Special Interest Group on Discourse and Dialogue",
month = sep,
year = "2024",
address = "Kyoto, Japan",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.sigdial-1.33",
doi = "10.18653/v1/2024.sigdial-1.33",
pages = "370--384",
}
License
This project is licensed under the Apache License, Version 2.0 (the "License"); you may not use the files except in compliance with the License. You may obtain a copy of the License at