CASA: Conversational Aspect Sentiment Analysis for Dialogue Understanding

Song, Linfeng; Xin, Chunlei; Lai, Su‐Yuan; Wang, Ante; Su, Jinsong; Xu, Kun

doi:10.1613/jair.1.12802

Cited by 10 publications

(3 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…7 Dialogue-level quadruple analysis Song et al (2022) introduced a new task called conversational aspect sentiment analysis (CASA), which aims to provide fine-grained sentiment information for dialogue understanding and planning.…”

Section: Summary Of the Best Methods For Each Approachmentioning

confidence: 99%

Exploring aspect-based sentiment quadruple extraction with implicit aspects, opinions, and ChatGPT: a comprehensive survey

Zhang,

Cheah,

Alyasiri

et al. 2024

Artif Intell Rev

View full text Add to dashboard Cite

In contrast to earlier ABSA studies primarily concentrating on individual sentiment components, recent research has ventured into more complex ABSA tasks encompassing multiple elements, including pair, triplet, and quadruple sentiment analysis. Quadruple sentiment analysis, also called aspect-category-opinion-sentiment quadruple Extraction (ACOSQE), aims to dissect aspect terms, aspect categories, opinion terms, and sentiment polarities while considering implicit sentiment within sentences. Nonetheless, a comprehensive overview of ACOSQE and its corresponding solutions is currently lacking. This is the precise gap that our survey seeks to address. To be more precise, we systematically reclassify all subtasks of ABSA, reorganizing existing research from the perspective of the involved sentiment elements, with a primary focus on the latest advancements in the ACOSQE task. Regarding solutions, our survey offers a comprehensive summary of the state-of-the-art utilization of language models within the ACOSQE task. Additionally, we explore the application of ChatGPT in sentiment analysis. Finally, we review emerging trends and discuss the challenges, providing insights into potential future directions for ACOSQE within the broader context of ABSA.

show abstract

Section: Summary Of the Best Methods For Each Approachmentioning

confidence: 99%

Exploring aspect-based sentiment quadruple extraction with implicit aspects, opinions, and ChatGPT: a comprehensive survey

Zhang,

Cheah,

Alyasiri

et al. 2024

Artif Intell Rev

View full text Add to dashboard Cite

show abstract

“…In recent years, effective dialogue agents have explored various knowledge for facilitating dialogue understanding and response generation, such as sentiment (Song et al, 2022), speaker emotion (Poria et al, 2019), discourse structure (Wang et al, 2021a(Wang et al, , 2023a, and external knowledge from different knowledge sources. This work falls into the last category of the above research lines, which has shown effectiveness in alleviating the hallucination problem.…”

Section: Knowledge-aided Dialogue Modelmentioning

confidence: 99%

Domain Adaptation for Conversational Query Production with the RAG Model Feedback

Wang,

Song,

et al. 2023

Findings of the Association for Computational Linguistics: EMNLP 2023

View full text Add to dashboard Cite

Conversational query production is an emerging fundamental task for the dialogue system, where search queries are generated to explore the vast and continually updating knowledge from a search engine. To accelerate this line of research, previous studies have released several datasets with human-annotated search queries. However, the limited annotations still can not cover conversations of various domains. To solve this challenge, we propose a novel domain adaptation framework. It is inspired by a weakly supervised learning algorithm from previous work (Wang et al., 2023b) that guides a model using reinforcement learning with BM25 scores as feedback. Though effective, it is fragile facing noisy content on webpages from a commercial search engine and variance in conversations because of ignoring deep semantic information of dialogue contexts. Thus, we improve the algorithm by taking the advance of retrieval-augmented generation (RAG) and exploring several practical techniques such as knowledge distillation for stable training. We conduct experiments in multiple settings across different languages. Guided by the RAG model feedback, our model is more robust and performs significantly better especially in a more challenging setting over strong baselines. 1

show abstract

“…Dialogue systems have attracted increasing attention from both academia and industry researches (Chen et al, 2017;Deriu et al, 2021;Gao et al, 2021a). The tasks can be commonly divided into two categories: task-oriented dialogue systems (Wen et al, 2017;Dinan et al, 2019;Mehri et al, 2020) and chit-chat dialogue systems (Ritter et al, 2011;Li et al, 2017;Cui et al, 2020;Song et al, 2022). The former aims to interact in the context of a specific task, while the latter chats with users without task and domain restrictions.…”

Section: Introductionmentioning

confidence: 99%

Semantic-based Pre-training for Dialogue Understanding

Bai¹,

Song²,

Zhang³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Pre-trained language models have made great progress on dialogue tasks. However, these models are typically trained on surface dialogue text, thus are proven to be weak in understanding the main semantic meaning of a dialogue context. We investigate Abstract Meaning Representation (AMR) as explicit semantic knowledge for pre-training models to capture the core semantic information in dialogues during pre-training. In particular, we propose a semantic-based pre-training framework that extends the standard pretraining framework (Devlin et al., 2019) by three tasks for learning 1) core semantic units, 2) semantic relations and 3) the overall semantic representation according to AMR graphs. Experiments on the understanding of both chit-chats and task-oriented dialogues show the superiority of our model. To our knowledge, we are the first to leverage a deep semantic representation for dialogue pretraining.

show abstract

CASA: Conversational Aspect Sentiment Analysis for Dialogue Understanding

Cited by 10 publications

References 30 publications

Exploring aspect-based sentiment quadruple extraction with implicit aspects, opinions, and ChatGPT: a comprehensive survey

Exploring aspect-based sentiment quadruple extraction with implicit aspects, opinions, and ChatGPT: a comprehensive survey

Domain Adaptation for Conversational Query Production with the RAG Model Feedback

Semantic-based Pre-training for Dialogue Understanding

Contact Info

Product

Resources

About