Towards Knowledge-Aware Document Systems: Modeling Semantic Coverage Relations via Answerability Detection
This work addresses the need for better information alignment across documents for tasks like retrieval and summarization, though it is incremental in improving existing methods.
The paper tackled the problem of modeling semantic coverage relations between documents by introducing a QA-based framework to classify document pairs into equivalence, inclusion, or semantic overlap, with results showing discriminative models like RoBERTa-base achieving up to 61.4% accuracy.
Understanding how information is shared across documents, regardless of the format in which it is expressed, is critical for tasks such as information retrieval, summarization, and content alignment. In this work, we introduce a novel framework for modelling Semantic Coverage Relations (SCR), which classifies document pairs based on how their informational content aligns. We define three core relation types: equivalence, where both texts convey the same information using different textual forms or styles; inclusion, where one document fully contains the information of another and adds more; and semantic overlap, where each document presents partially overlapping content. To capture these relations, we adopt a question answering (QA)-based approach, using the answerability of shared questions across documents as an indicator of semantic coverage. We construct a synthetic dataset derived from the SQuAD corpus by paraphrasing source passages and selectively omitting information, enabling precise control over content overlap. This dataset allows us to benchmark generative language models and train transformer-based classifiers for SCR prediction. Our findings demonstrate that discriminative models significantly outperform generative approaches, with the RoBERTa-base model achieving the highest accuracy of 61.4% and the Random Forest-based model showing the best balance with a macro-F1 score of 52.9%. The results show that QA provides an effective lens for assessing semantic relations across stylistically diverse texts, offering insights into the capacity of current models to reason about information beyond surface similarity. The dataset and code developed in this study are publicly available to support reproducibility.