Deep Content Understanding Toward Entity and Aspect Target Sentiment Analysis on Foundation Models

TL;DR

The paper introduces EASTE, a fine-grained ABSA task that extracts $(t_i, e_i, a_i, s_i)$ triplets by separately identifying entities and aspects. It develops a unified-loss token-classification method and explores text-generation pipelines with foundation models under zero-/few-shot and PEFT setups, evaluated on the SemEval-2016 Restaurant dataset. The results show state-of-the-art performance with full fine-tuning on Flan-T5-XL (F1 $= 76.38$), while prompting-based LLMs offer strong recall with varying precision, and PEFT methods provide competitive alternatives. This work demonstrates effective deep content understanding for complex sentiment analysis and offers practical guidance on model selection and adaptation for ABSA tasks.

Abstract

Introducing Entity-Aspect Sentiment Triplet Extraction (EASTE), a novel Aspect-Based Sentiment Analysis (ABSA) task which extends Target-Aspect-Sentiment Detection (TASD) by separating aspect categories (e.g., food#quality) into pre-defined entities (e.g., meal, drink) and aspects (e.g., taste, freshness) which add a fine-gainer level of complexity, yet help exposing true sentiment of chained aspect to its entity. We explore the task of EASTE solving capabilities of language models based on transformers architecture from our proposed unified-loss approach via token classification task using BERT architecture to text generative models such as Flan-T5, Flan-Ul2 to Llama2, Llama3 and Mixtral employing different alignment techniques such as zero/few-shot learning, Parameter Efficient Fine Tuning (PEFT) such as Low-Rank Adaptation (LoRA). The model performances are evaluated on the SamEval-2016 benchmark dataset representing the fair comparison to existing works. Our research not only aims to achieve high performance on the EASTE task but also investigates the impact of model size, type, and adaptation techniques on task performance. Ultimately, we provide detailed insights and achieving state-of-the-art results in complex sentiment analysis.

Figures (1)

• Figure 1: Unified-loss architecture based on BERT token classification where each loss function of $entity, aspect, sentiment$ are weighted thoroughly.