Multi-Perspective Fusion Network for Commonsense Reading Comprehension

Abstract

Commonsense Reading Comprehension (CRC) is a significantly challenging task, aiming at choosing the right answer for the question referring to a narrative passage, which may require commonsense knowledge inference. Most of the existing approaches only fuse the interaction information of choice, passage, and question in a simple combination manner from a union perspective, which lacks the comparison information on a deeper level. Instead, we propose a Multi-Perspective Fusion Network (MPFN), extending the single fusion method with multiple perspectives by introducing the difference and similarity fusion. More comprehensive and accurate information can be captured through the three types of fusion. We design several groups of experiments on MCScript dataset [11] to evaluate the effectiveness of the three types of fusion respectively. From the experimental results, we can conclude that the difference fusion is comparable with union fusion, and the similarity fusion needs to be activated by the union fusion. The experimental result also shows that our MPFN model achieves the state-of-the-art with an accuracy of 83.52% on the official test set.

A Survey on Machine Reading Comprehension Systems

“…2018a; Liu et al . 2018a). In MRC systems, like other NLP tasks, these architectures have been commonly used in different parts of the pipeline, such as for representing questions and contexts (Chen et al .…”

“…In MRC systems, like other NLP tasks, these architectures are used to represent text data in different parts of their pipelines, such as for representing the questions and passages. Bidirectional versions of LSTM [15][16][17][18][19][20] and GRU [21][22][23][24][25][26][27][28] are also very popular in this task. LSTM and GRU are also used in higher levels of the MRC system architecture like in the modeling layer [18,19,[29][30][31].…”

“…2018a; Liu et al . 2018a) or in higher levels of the MRC system such as the modeling layer (Choi et al . 2017b; Seo et al .…”