# On the Challenges of Evaluating Compositional Explanations in Multi-Hop Inference: Relevance, Completeness, and Expert Ratings

arXiv: https://arxiv.org/abs/2109.03334 · HTML: https://mlsanity.com/paper/2109.03334/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 41 references that are in the map.
- First posted on arXiv 7 Sep 2021.

## Builds on (most cited first, 22 of 22)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering](https://mlsanity.com/paper/1809.02789/) (2018)
- [MS MARCO: A Human Generated MAchine Reading COmprehension Dataset](https://mlsanity.com/paper/1611.09268/) (2016)
- [RACE: Large-scale ReAding Comprehension Dataset From Examinations](https://mlsanity.com/paper/1704.04683/) (2017)
- [UnifiedQA: Crossing Format Boundaries With a Single QA System](https://mlsanity.com/paper/2005.00700/) (2020)
- [Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translation](https://mlsanity.com/paper/2104.14478/) (2021)
- [Transformers as Soft Reasoners over Language](https://mlsanity.com/paper/2002.05867/) (2020)
- [QASC: A Dataset for Question Answering via Sentence Composition](https://mlsanity.com/paper/1910.11473/) (2019)
- [KagNet: Knowledge-Aware Graph Networks for Commonsense Reasoning](https://mlsanity.com/paper/1909.02151/) (2019)
- [WorldTree: A Corpus of Explanation Graphs for Elementary Science Questions supporting Multi-Hop Inference](https://mlsanity.com/paper/1802.03052/) (2018)
- [Learning to Explain: Datasets and Models for Identifying Valid Reasoning Chains in Multihop Question-Answering](https://mlsanity.com/paper/2010.03274/) (2020)
- [NLProlog: Reasoning with Weak Unification for Question Answering in Natural Language](https://mlsanity.com/paper/1906.06187/) (2019)
- [R4C: A Benchmark for Evaluating RC Systems to Get the Right Answer for the Right Reason](https://mlsanity.com/paper/1910.04601/) (2019)
- [Learning-to-Rank with BERT in TF-Ranking](https://mlsanity.com/paper/2004.08476/) (2020)
- [A Survey on Explainability in Machine Reading Comprehension](https://mlsanity.com/paper/2010.00389/) (2020)
- [TF-Ranking: Scalable TensorFlow Library for Learning-to-Rank](https://mlsanity.com/paper/1812.00073/) (2018)
- [Question Answering via Integer Programming over Semi-Structured Knowledge](https://mlsanity.com/paper/1604.06076/) (2016)
- [Autoregressive Reasoning over Chains of Facts with Transformers](https://mlsanity.com/paper/2012.11321/) (2020)

## Built on by (most cited first, 5 of 5)

- [ScienceWorld: Is your Agent Smarter than a 5th Grader?](https://mlsanity.com/paper/2203.07540/) (2022)
- [Knowledge-enhanced Agents for Interactive Text Games](https://mlsanity.com/paper/2305.05091/) (2023)
- [Scientific Explanation and Natural Language: A Unified Epistemological-Linguistic Perspective for Explainable AI](https://mlsanity.com/paper/2205.01809/) (2022)
- [From Models to Microtheories: Distilling a Model's Topical Knowledge for Grounded Question Answering](https://mlsanity.com/paper/2412.17701/) (2024)
- [Generating Data-Driven Reasoning Rubrics for Domain-Adaptive Reward Modeling](https://mlsanity.com/paper/2602.06795/) (2026)

## Abstract (arXiv)

Building compositional explanations requires models to combine two or more facts that, together, describe why the answer to a question is correct. Typically, these "multi-hop" explanations are evaluated relative to one (or a small number of) gold explanations. In this work, we show these evaluations substantially underestimate model performance, both in terms of the relevance of included facts, as well as the completeness of model-generated explanations, because models regularly discover and produce valid explanations that are different than gold explanations. To address this, we construct a large corpus of 126k domain-expert (science teacher) relevance ratings that augment a corpus of explanations to standardized science exam questions, discovering 80k additional relevant facts not rated as gold. We build three strong models based on different methodologies (generation, ranking, and schemas), and empirically show that while expert-augmented ratings provide better estimates of explanation quality, both original (gold) and expert-augmented automatic evaluations still substantially underestimate performance by up to 36% when compared with full manual expert judgements, with different models being disproportionately affected. This poses a significant methodological challenge to accurately evaluating explanations produced by compositional reasoning models.