PortNLP / MEEP

Code for Prompting Large Language Models for Dialogue Evaluation in Multilingual Settings

MEEP

Metric for Evaluation of Engagingness using Prompting

This repository contains the code for dialogue evaluation in multilingual settings that was used to produce the tests in the paper

MEEP: Is this Engaging? Prompting Large Language Models for Dialogue Evaluation in Multilingual Settings

published in EMNLP 2023 Findings.

Directory Structure

The directory structure should include a Datasets directory at the same level as the Evaluation directory.

• MEEP
• |___ Datasets
• |___ DSTC_11_Track_4
• |___ ...
• .
• .
• .
• |___ Evaluation
• |___ preprocessors

Prompts

The prompts used in the paper are in engagingnessprompt.py.

Datasets

Download the datasets from DSTC11 into the Datasets directory. Include the DSTC_11_Track_4 directory within the Datasets directory.

SEE dataset

Code for random selection of samples from the SEE dataset is in see_selector.py. This is to show our process, but will create a different dataset than the one we used. To recreate our dataset, use see_subset_maker.py. Use the same file structure found in the DSTC11 data.

Citation

If you use the material in this repository or in the paper, please cite our work as

Amila Ferron, Amber Shore, Ekata Mitra, and Ameeta Agrawal. 2023. MEEP: Is this Engaging? Prompting Large Language Models for Dialogue Evaluation in Multilingual Settings. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 2078–2100, Singapore. Association for Computational Linguistics.

Bibtex and other citation code can be found on the paper's ACL Anthology page.