# Multi-hop Inference for Sentence-level TextGraphs: How Challenging is Meaningfully Combining Information for Science Question Answering?

arXiv: https://arxiv.org/abs/1805.11267 · HTML: https://mlsanity.com/paper/1805.11267/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 May 2018.

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

- [Rationalization for Explainable NLP: A Survey](https://mlsanity.com/paper/2301.08912/) (2023)
- [Team Enigma at ArgMining-EMNLP 2021: Leveraging Pre-trained Language Models for Key Point Matching](https://mlsanity.com/paper/2110.12370/) (2021)
- [Diff-Explainer: Differentiable Convex Optimization for Explainable Multi-hop Inference](https://mlsanity.com/paper/2105.03417/) (2021)
- [Exploring Equation as a Better Intermediate Meaning Representation for Numerical Reasoning](https://mlsanity.com/paper/2308.10585/) (2023)
- [Explainable Compliance Detection with Multi-Hop Natural Language Inference on Assurance Case Structure](https://mlsanity.com/paper/2506.08713/) (2025)

## Abstract (arXiv)

Question Answering for complex questions is often modeled as a graph construction or traversal task, where a solver must build or traverse a graph of facts that answer and explain a given question. This "multi-hop" inference has been shown to be extremely challenging, with few models able to aggregate more than two facts before being overwhelmed by "semantic drift", or the tendency for long chains of facts to quickly drift off topic. This is a major barrier to current inference models, as even elementary science questions require an average of 4 to 6 facts to answer and explain. In this work we empirically characterize the difficulty of building or traversing a graph of sentences connected by lexical overlap, by evaluating chance sentence aggregation quality through 9,784 manually-annotated judgments across knowledge graphs built from three free-text corpora (including study guides and Simple Wikipedia). We demonstrate semantic drift tends to be high and aggregation quality low, at between 0.04% and 3%, and highlight scenarios that maximize the likelihood of meaningfully combining information.