Multi-Label Few-Shot Learning for Aspect Category Detection

Hu, Mengting; Zhao, Shiwan; Guo, Honglei; Xue, Chao; Gao, Hang; Gao, Tiegang; Cheng, Renhong; Su, Zhong

doi:10.18653/v1/2021.acl-long.495

Cited by 26 publications

(19 citation statements)

References 26 publications

(25 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As shown in Figure 1, as "food" is the target aspect, "staff" will be treated as a noise aspect for the sentence "It is the staff and food quality that really needs fixing." Although previous works (Hu et al 2021;Zhao et al 2022;Liu et al 2022) exploit the aspect information to guide an attention mechanism to alleviate this issue, it is hard to ensure that novel aspects can establish accurate attention associations with sentence features during inference. Secondly, existing FS-ACD methods consistently follow the prototypical network (Snell et al 2017), which learns a prototype for each aspect and uses the distance between query samples and prototypes to predict labels.…”

Section: Hotelmentioning

confidence: 99%

See 1 more Smart Citation

Variational Hybrid-Attention Framework for Multi-Label Few-Shot Aspect Category Detection

Peng,

Chen,

Shou

et al. 2024

AAAI

View full text Add to dashboard Cite

Multi-label few-shot aspect category detection (FS-ACD) is a challenging sentiment analysis task, which aims to learn a multi-label learning paradigm with limited training data. The difficulty of this task is how to use limited data to generalize effective discriminative representations for different categories. Nowadays, all advanced FS-ACD works utilize the prototypical network to learn label prototypes to represent different aspects. However, such point-based estimation methods are inherently noise-susceptible and bias-vulnerable. To this end, this paper proposes a novel Variational Hybrid-Attention Framework (VHAF) for the FS-ACD task. Specifically, to alleviate the data noise, we adopt a hybrid-attention mechanism to generate more discriminative aspect-specific embeddings. Then, based on these embeddings, we introduce the variational distribution inference to obtain the aspect-specific distribution as a more robust aspect representation, which can eliminate the scarce data bias for better inference. Moreover, we further leverage an adaptive threshold estimation to help VHAF better identify multiple relevant aspects. Extensive experiments on three datasets demonstrate the effectiveness of our VHAF over other state-of-the-art methods. Code is available at https://github.com/chengzju/VHAF.

show abstract

Section: Hotelmentioning

confidence: 99%

“…FSL is a human-like learning paradigm that can quickly generalize novel classes with a few training data by exploiting prior knowledge. The pioneering work (Hu et al 2021) is Support Set hotel (1) This hotel is terrible with even worse service.…”

Section: Introductionmentioning

confidence: 99%

Variational Hybrid-Attention Framework for Multi-Label Few-Shot Aspect Category Detection

Peng,

Chen,

Shou

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…In this example, the aspect category is "battery." ACD methods can be classified into two types by Zhang et al (2022): supervised (Zhou et al, 2015;Movahedi et al, 2019;Ghadery et al, 2019;Hu et al, 2021) and unsupervised methods (Tulkens and van Cranenburgh, 2020;Shi et al, 2021) based on the availability of labeled data.…”

Section: Aspect Category Detection (Acd)mentioning

confidence: 99%

“…To address the challenge in ASPE, researchers have developed various approaches, including pipeline methods, unified tagging schemas (Mitchell et al, 2013;Zhang et al, 2015;, multi-task learning (Hu et al, 2021;Chen and Qian, 2020b), and span-based (Hu et al, 2019a) techniques. In recent works, additional methods such as few-shot learning (Hosseini-Asl et al, 2022), zero-shot ABSA (Shu et al, 2022) crosslingual ABSA (Zhang et al, 2021b), Machine Reading Comprehension (MRC) (Yu et al, 2021), and structured sentiment analysis (dependency graph parsing) (Barnes et al, 2021) have been explored to improve performance.…”

Section: Aspect Sentiment Pair Extraction (Aspe)mentioning

confidence: 99%

A Survey on Aspect-Based Sentiment Quadruple Extraction with Implicit Aspects and Opinions

Zhang

Cheah

Alyasiri

et al. 2023

Preprint

View full text Add to dashboard Cite

Quadruple sentiment analysis is an essential task in aspect-based sentiment analysis that aims to understand people's viewpoints by analyzing four elements in text: category, aspect, opinion, and sentiment polarity. In recent years, quadruple sentiment analysis tasks have received extensive research attention for capturing comprehensive sentiment information in sentences. This paper focuses on the latest progress in quadruple sentiment analysis tasks. Additionally, the paper provides a summary of the application of methods based on pre-trained language models and datasets in quadruple sentiment analysis with implicit aspects and opinions. It discusses techniques for constructing more practical quadruple sentiment analysis systems in generative large model scenarios. The paper's final section outlines quadruple sentiment analysis research's current challenges and future directions. Moreover, the paper suggests that using advanced language models, such as ChatGPT, holds promise for improving the accuracy and efficiency of quadruple sentiment analysis. Overall, this paper sheds light on the importance of quadruple sentiment analysis and highlights opportunities for further research in this field.

show abstract

“…Zhang et al [45] treated the ATE task as a question-answering task. Hu et al [46] viewed the aspect extraction task as a multi-label learning problem. They employed the prototypical network to develop a few-shot learning model and designed two attention mechanisms to alleviate the noise.…”

Section: Related Workmentioning

confidence: 99%

Aspect term extraction via information-augmented neural network

Liu

Shen

2022

Complex Intell. Syst.

View full text Add to dashboard Cite

Aspect term extraction (ATE) aims at identifying the aspect terms that are expressed in a sentence. Recently, Seq2Seq learning has been employed in ATE and significantly improved performance. However, it suffers from some weaknesses, such as lacking the ability to encode the more informative information and integrate information of surrounding words in the encoder. The static word embeddings employed in ATE fall short of modeling the dynamic meaning of words. To alleviate the problems mentioned above, this paper proposes the information-augmented neural network (IANN) which is a novel Seq2Seq learning framework. In IANN, a specialized neural network is developed as the key module of the encoder, named multiple convolution with recurrence network (MCRN), to encode the more informative information and integrate information of surrounding words in the encoder. The contextualized embedding layer is designed to capture the dynamic word sense. Besides, the novel AO ({Aspect, Outside}) tags are proposed as the less challenging tagging scheme. A lot of experiments have been performed on three widely used datasets. These experiments demonstrate that the proposed IANN acquires state-of-the-art results and validate that the proposed IANN is a powerful method for the ATE task.

show abstract

Multi-Label Few-Shot Learning for Aspect Category Detection

Cited by 26 publications

References 26 publications

Variational Hybrid-Attention Framework for Multi-Label Few-Shot Aspect Category Detection

Variational Hybrid-Attention Framework for Multi-Label Few-Shot Aspect Category Detection

A Survey on Aspect-Based Sentiment Quadruple Extraction with Implicit Aspects and Opinions

Aspect term extraction via information-augmented neural network

Contact Info

Product

Resources

About