Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Semantic Integration of Opinions: Unified Opinion Concepts Ontology and Extraction Task

Gaurav Negi, Dhairya Dalal, Omnia Zayed, Paul Buitelaar

TL;DR

This work introduces the Unified Opinion Concepts (UOC) ontology to semantically unify opinion representations across NLP paradigms and semantic-web ontology. It then defines Unified Opinion Concept Extraction (UOCE), an NLP task that extracts exhaustive opinion tuples grounded in UOC, supported by a semantically validated, extended evaluation dataset and tailored metrics. Baseline experiments using diverse large language models and two prompt styles (NLPrompt and OntoPrompt) demonstrate expressivity gains over existing fine-grained opinion mining methods such as ACOS and ASTE, while also revealing challenges in capturing qualifiers and reasons. The study provides a pathway toward interoperable, semantically rich opinion extraction with practical implications for cross-domain knowledge graphs and semantically aware NLP systems.

Abstract

This paper introduces the Unified Opinion Concepts (UOC) ontology to integrate opinions within their semantic context. The UOC ontology bridges the gap between the semantic representation of opinion across different formulations. It is a unified conceptualisation based on the facets of opinions studied extensively in NLP and semantic structures described through symbolic descriptions. We further propose the Unified Opinion Concept Extraction (UOCE) task of extracting opinions from the text with enhanced expressivity. Additionally, we provide a manually extended and re-annotated evaluation dataset for this task and tailored evaluation metrics to assess the adherence of extracted opinions to UOC semantics. Finally, we establish baseline performance for the UOCE task using state-of-the-art generative models.

Towards Semantic Integration of Opinions: Unified Opinion Concepts Ontology and Extraction Task

TL;DR

This work introduces the Unified Opinion Concepts (UOC) ontology to semantically unify opinion representations across NLP paradigms and semantic-web ontology. It then defines Unified Opinion Concept Extraction (UOCE), an NLP task that extracts exhaustive opinion tuples grounded in UOC, supported by a semantically validated, extended evaluation dataset and tailored metrics. Baseline experiments using diverse large language models and two prompt styles (NLPrompt and OntoPrompt) demonstrate expressivity gains over existing fine-grained opinion mining methods such as ACOS and ASTE, while also revealing challenges in capturing qualifiers and reasons. The study provides a pathway toward interoperable, semantically rich opinion extraction with practical implications for cross-domain knowledge graphs and semantically aware NLP systems.

Abstract

This paper introduces the Unified Opinion Concepts (UOC) ontology to integrate opinions within their semantic context. The UOC ontology bridges the gap between the semantic representation of opinion across different formulations. It is a unified conceptualisation based on the facets of opinions studied extensively in NLP and semantic structures described through symbolic descriptions. We further propose the Unified Opinion Concept Extraction (UOCE) task of extracting opinions from the text with enhanced expressivity. Additionally, we provide a manually extended and re-annotated evaluation dataset for this task and tailored evaluation metrics to assess the adherence of extracted opinions to UOC semantics. Finally, we establish baseline performance for the UOCE task using state-of-the-art generative models.

Paper Structure

This paper contains 27 sections, 5 equations, 3 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Ontological Methods.
  4. Neuro-Symbolic Methods.
  5. Unified Opinion Concept Ontology
  6. Modelling Ontological Concepts and Relationships
  7. Unified Opinion Concept Extraction (UOCE)
  8. Problem Definition
  9. Evaluation Metrics
  10. Tuple-Level Exact Match Metric
  11. Component-Level Exact Match Metric
  12. Dataset
  13. Semantic Data Validation
  14. Evaluation Dataset Creation
  15. Baseline Methods
  16. ...and 12 more sections

Figures (3)

  • Figure 1: Structured Sentiment Analysis
  • Figure 2: Unified Opinions Concepts (UOC) Ontology Diagram
  • Figure 3: UOC Sentiment extracted from: "I had hoped for better battery life , as it had only about 2-1/2 hours doing heavy computations (8 threads using 100 % of the CPU)"