Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Make Compound Sentences Simple to Analyze: Learning to Split Sentences for Aspect-based Sentiment Analysis

Yongsik Seo, Sungwon Song, Ryang Heo, Jieyong Kim, Dongha Lee

TL;DR

This paper proposes Aspect Term Oriented Sentence Splitter (ATOSS), which simplifies compound sentence into simpler and clearer forms, thereby clarifying their structure and intent and outperforms existing methods in both ASQP and ACOS tasks.

Abstract

In the domain of Aspect-Based Sentiment Analysis (ABSA), generative methods have shown promising results and achieved substantial advancements. However, despite these advancements, the tasks of extracting sentiment quadruplets, which capture the nuanced sentiment expressions within a sentence, remain significant challenges. In particular, compound sentences can potentially contain multiple quadruplets, making the extraction task increasingly difficult as sentence complexity grows. To address this issue, we are focusing on simplifying sentence structures to facilitate the easier recognition of these elements and crafting a model that integrates seamlessly with various ABSA tasks. In this paper, we propose Aspect Term Oriented Sentence Splitter (ATOSS), which simplifies compound sentence into simpler and clearer forms, thereby clarifying their structure and intent. As a plug-and-play module, this approach retains the parameters of the ABSA model while making it easier to identify essential intent within input sentences. Extensive experimental results show that utilizing ATOSS outperforms existing methods in both ASQP and ACOS tasks, which are the primary tasks for extracting sentiment quadruplets.

Make Compound Sentences Simple to Analyze: Learning to Split Sentences for Aspect-based Sentiment Analysis

TL;DR

This paper proposes Aspect Term Oriented Sentence Splitter (ATOSS), which simplifies compound sentence into simpler and clearer forms, thereby clarifying their structure and intent and outperforms existing methods in both ASQP and ACOS tasks.

Abstract

In the domain of Aspect-Based Sentiment Analysis (ABSA), generative methods have shown promising results and achieved substantial advancements. However, despite these advancements, the tasks of extracting sentiment quadruplets, which capture the nuanced sentiment expressions within a sentence, remain significant challenges. In particular, compound sentences can potentially contain multiple quadruplets, making the extraction task increasingly difficult as sentence complexity grows. To address this issue, we are focusing on simplifying sentence structures to facilitate the easier recognition of these elements and crafting a model that integrates seamlessly with various ABSA tasks. In this paper, we propose Aspect Term Oriented Sentence Splitter (ATOSS), which simplifies compound sentence into simpler and clearer forms, thereby clarifying their structure and intent. As a plug-and-play module, this approach retains the parameters of the ABSA model while making it easier to identify essential intent within input sentences. Extensive experimental results show that utilizing ATOSS outperforms existing methods in both ASQP and ACOS tasks, which are the primary tasks for extracting sentiment quadruplets.
Paper Structure (38 sections, 2 equations, 4 figures, 8 tables)

This paper contains 38 sections, 2 equations, 4 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Sentiment Quadruplet Prediction
  4. Distillation of LLM's Reasoning Ability
  5. Preliminaries
  6. Problem Formulation
  7. Analysis of ABSA Performance based on Sentence Structural Complexity
  8. Methodology
  9. Aspect-Oriented Splitting Strategy
  10. Distillation of LLM's Splitting Ability
  11. Split sentence generation
  12. Split sentence selection
  13. Supervised fine-tuning (SFT)
  14. Alignment with Sentence Preference
  15. Preferred sentence selection
  16. ...and 23 more sections

Figures (4)

  • Figure 1: Existing ABSA models struggle to accurately predict quadruplets in sentence with compound syntactic structures but perform well when the sentences are provided in simpler and clearer forms.
  • Figure 2: Performance changes of ABSA models (Left: GPT-4-turbo, Right: MvP) w.r.t. the number of candidate split sentences. (Task: ACOS, Dataset: Rest16)
  • Figure 3: Overall framework for training and utilizing AToss for ABSA tasks. The training process involves (1) distillation of LLM's capability for sentence splitting, and (2) alignment with a target model's sentence preference. The inference process (3) predicts the quadruplets by taking sentences split by AToss as the input. AToss, as a plug-and-play module, can enhance prediction accuracy without requiring updates to the target model’s parameters.
  • Figure 4: Left: Model: MvP, Task: ACOS, Dataset: Rest16. Right: F1 improvement when each model is trained on split sentences instead of original sentences.