Explanation Faithfulness Evaluation Measuress

A website to track the progress of the Explanation Faithfulness Evaluation Measures project.

Overview

As artificial intelligence systems, especially large language models, gain popularity there is an increased desire to trust these systems. One common method to increase trust is to provide explanations. These explanations provide information about the system’s workings and the knowledge used in its general reasoning processes or the processes behind a specific decision [1]. However, ensuring that these explanations are faithful, that they accurately represent that reasoning process, is a difficult and ongoing task [2]. Over 60 measures have been proposed since 2020 to quantify how faithful an explanation is, but it is difficult to compare them to decide which one is most appropriate for a given use case.

This project aims to develop structured representation of these measures and provide a centralized resource for researchers to use. Currently we have released the beta version of the EFEMO ontology and are developing an automatic extraction pipeline for contructing a knowledge graph. Please check back frequently for updates!

[1] Chari, S., Gruen, D. M., Seneviratne, O., & McGuinness, D. L. (2020). Foundations of explainable knowledge-enabled systems. In Knowledge Graphs for eXplainable Artificial Intelligence: Foundations, Applications and Challenges (pp. 23-48). IOS Press.

[2] Jacovi, A., & Goldberg, Y. (2020). Towards faithfully interpretable NLP systems: How should we define and evaluate faithfulness?. arXiv preprint arXiv:2004.03685.

List of Resources

Resources	Links
Ontology Project	EFEMO
Knowledge Grph Project	EFEMO-KG
Papers and Posters	Publications

Team

Danielle Villa*¹, Maria Chang² , Deborah L. McGuinness¹

¹Rensselaer Polytechnic Institute | ²IBM Research

*Contact at: villad4@rpi.edu