A T5-based interpretable reading comprehension model with more accurate evidence training

Article Computer Science, Information Systems

Semantic matching in machine reading comprehension: An empirical study

Qian Liu et al.

Summary: This paper conducts a systematic empirical study on semantic matching in machine reading comprehension (MRC). It formulates a two-stage framework and compares different setups of semantic matching modules on four MRC datasets. The study finds that semantic matching improves the effectiveness and efficiency of MRC, especially for answering questions with noisy and adversarial context. Matching coarse-grained context to questions and using semantic matching modules is more effective than fine-grained context matching, such as sentences and spans. However, semantic matching decreases the performance on why questions, suggesting that it is more helpful for questions that can be answered by retrieving information from a single sentence.

INFORMATION PROCESSING & MANAGEMENT (2023)

添加到收藏夹

Article Computer Science, Interdisciplinary Applications

MRC4BioER: Joint extraction of biomedical entities and relations in the machine reading comprehension framework

Cong Sun et al.

Summary: By introducing a novel tagging strategy and Query in the machine reading comprehension framework, the method successfully achieves joint extraction of overlapping triplets in the biomedical domain, enhancing the model's ability to deal with such complex relationships.

JOURNAL OF BIOMEDICAL INFORMATICS (2022)

添加到收藏夹

Article Multidisciplinary Sciences

ExpMRC: explainability evaluation for machine reading comprehension

Yiming Cui et al.

Summary: This paper proposes a new benchmark called ExpMRC for evaluating the textual explainability of Machine Reading Comprehension (MRC) systems. The benchmark consists of four subsets, with additional annotations of answer evidence. State-of-the-art pre-trained language models are used to build baseline systems, and unsupervised approaches are adopted to extract answer and evidence spans. Experimental results show that the current systems still have a gap from human performance.

HELIYON (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

A pattern-first pipeline approach for entity and relation extraction

Zheng Chen et al.

Summary: Entity-relation extraction is the task of extracting entities and their semantic relations from unstructured text. Recent studies have shown that Machine Reading Comprehension (MRC) based methods achieve significant results in this task. However, traditional entity-first methods suffer from entity redundancy and error propagation. To address these issues, we propose a pattern-first pipeline approach.

NEUROCOMPUTING (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Sentence Similarity Based on Contexts

Xiaofei Sun et al.

Summary: This study proposes a new framework to address two challenges in measuring sentence similarity. The framework is based on the core idea that the meaning of a sentence should be defined by its contexts and measures sentence similarity by comparing the probabilities of generating two sentences given the same context. Experimental results show significant performance improvements over existing baselines in both supervised and unsupervised settings.

TRANSACTIONS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Enhancing Transformer-based language models with commonsense representations for knowledge-driven machine comprehension

Ronghan Li et al.

Summary: The paper introduces three injection methods and a mask mechanism for fine-tuning knowledge-driven MRC tasks, achieving significant performance improvements and demonstrating the effectiveness and robustness of the incremental model.

KNOWLEDGE-BASED SYSTEMS (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence