Argument Pair Extraction via Attention-guided Multi-Layer Multi-Cross Encoding

Abstract

Argument pair extraction (APE) is a research task for extracting arguments from two passages and identifying potential argument pairs. Prior research work treats this task as a sequence labeling problem and a binary classification problem on two passages that are directly concatenated together, which has a limitation of not fully utilizing the unique characteristics and inherent relations of two different passages. This paper proposes a novel attention-guided multi-layer multi-cross encoding scheme to address the challenges. The new model processes two passages with two individual sequence encoders and updates their representations using each other's representations through attention. In addition, the pair prediction part is formulated as a tablefilling problem by updating the representations of two sequences' Cartesian product. Furthermore, an auxiliary attention loss is introduced to guide each argument to align to its paired argument. An extensive set of experiments show that the new model significantly improves the APE performance over several alternatives 1 .

IAM: A Comprehensive and Large-Scale Dataset for Integrated Argument Mining Tasks

“…In recent years, there is a tremendous amount of research effort in the computational argumentation research field Bar-Haim et al, 2021), such as argument components identification Rinott et al, 2015;Lippi and Torroni, 2016;, argument classification and clustering (Reimers et al, 2019), argument relation prediction (Boltužić and Šnajder, 2016;Chakrabarty et al, 2019), argument pair extraction (Cabrio and Villata, 2012;Cheng et al, 2020Cheng et al, , 2021, argument quality assessment (Habernal and Gurevych, 2016;Wachsmuth et al, 2017;Gretz et al, 2020;Toledo et al, 2019), listening comprehension (Mirkin et al, 2018), etc. Meanwhile, researchers have been exploring new datasets and methods to automate the debating preparation process, such as project debater (Slonim et al, 2021), etc.…”

“…We implement the sentence-pair classification model and the multi-label model for CESC with the aid of SimpleTransformers (Rajapakse, 2019). The multi-task model for CEPE is based on the implementation of the multi-task framework by Cheng et al (2021). All models are run with V100 GPU.…”

“…Table 13 shows the effect of using different numbers of layers in the multi-task model. More model details regarding each layer could be found in (Cheng et al, 2021)…”

Have my arguments been replied to? Argument Pair Extraction as Machine Reading Comprehension

“…As a salient part of argument mining (AM), the analysis of dialogical argumentation has received increasing research attention (Morio and Fujita, 2018;Chakrabarty et al, 2019;Cheng et al, 2021;Yuan et al, 2021). Argument pair extraction (APE), proposed by Cheng et al (2020), is a new task within this field that focuses on extracting interactive argument pairs from two interrelated documents (e.g., peer reviewer and rebuttal).…”

“…Previous works (Cheng et al, 2020(Cheng et al, , 2021 commonly address APE by decomposing it into two sentence-level subtasks, i.e., a sequence labeling task and a sentence relation classification task. These methods identify arguments by sentencelevel sequence labeling and determine whether two sentences belong to the same argument pair by sentence relation classification.…”