First Target and Opinion then Polarity: Enhancing Target-opinion Correlation for Aspect Sentiment Triplet Extraction

Huang, Lianzhe; Wang, Peiyi; Li, Sujian; Li, Tianyu; Zhang, Xiaodong; Cheng, Zhicong; Yin, Dawei; Wang, Houfeng

doi:10.48550/arxiv.2102.08549

Cited by 7 publications

(14 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…5 As shown in Table VI, firstly, Both, OnlyTab and [14] outperforms OnlySeq and GTS, which shows the importance of compositional feature for ASTE. Secondly, our OnlyTab model outperforms [14] on three of four datasets. which shows our table encoder is relatively better to capture the compositional feature than [14].…”

Section: B Effect Of Table Encodermentioning

confidence: 91%

“…Secondly, our OnlyTab model outperforms [14] on three of four datasets. which shows our table encoder is relatively better to capture the compositional feature than [14]. In addition, we also conduct a case study to intuitively show the effect of our table encoder, as shown in Figure 7, the top-right and bottom-left triangles of the matrix represent the sentiment classification results of OnlyTab and OnlySeq, respectively.…”

Section: B Effect Of Table Encodermentioning

confidence: 91%

“…The way RINANTE+ determine the correctness of a triplet is the same as CMLA+. [3], [15], [16], [13], [5], [12] and [14], respectively. The best results on each dataset are in boldface.…”

Section: B Baselinesmentioning

confidence: 99%

“…Compared with ASPE, ASTE can give reasons for sentiment classification, so it is more interpretable. In parallel with our work, [14] inserts four special tags into sentence to guide the encoder to extract compositional feature. [12] uses the machine reading comprehension framework, and propose a query such as 'whats the sentiment polarity of [price] and [low]' to classify the sentiment.…”

Section: Related Workmentioning

confidence: 99%

“…The F1 results of full model, models using only one encoder, GTS and[14] on four datasets of ASTE-DATA-V2.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Explicit Interaction Network for Aspect Sentiment Triplet Extraction

Wang¹,

Huang²,

Li³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Aspect Sentiment Triplet Extraction (ASTE) aims to recognize targets, their sentiment polarities and opinions explaining the sentiment from a sentence. ASTE could be naturally divided into 3 atom subtasks, namely target detection, opinion detection and sentiment classification. We argue that the proper subtask combination, compositional feature extraction for targetopinion pairs, and interaction between subtasks would be the key to success. Prior work, however, may fail on 'one-to-many' or 'many-to-one' situations, or derive non-existent sentiment triplets due to defective subtask formulation, sub-optimal feature representation or the lack of subtask interaction. In this paper, we divide ASTE into target-opinion joint detection and sentiment classification subtasks, which is in line with human cognition, and correspondingly propose sequence encoder and table encoder. Table encoder extracts sentiment at token-pair level, so that the compositional feature between targets and opinions can be easily captured. To establish explicit interaction between subtasks, we utilize the table representation to guide the sequence encoding, and inject the sequence features back into the table encoder. Experiments show that our model outperforms state-of-the-art methods on six popular ASTE datasets.

show abstract

Section: B Effect Of Table Encodermentioning

confidence: 91%

Section: B Effect Of Table Encodermentioning

confidence: 91%

“…The way RINANTE+ determine the correctness of a triplet is the same as CMLA+. [3], [15], [16], [13], [5], [12] and [14], respectively. The best results on each dataset are in boldface.…”

Section: B Baselinesmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

“…The F1 results of full model, models using only one encoder, GTS and[14] on four datasets of ASTE-DATA-V2.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Explicit Interaction Network for Aspect Sentiment Triplet Extraction

Wang¹,

Huang²,

Li³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

SpanMTL: a span-based multi-table labeling for aspect-oriented fine-grained opinion extraction

Xing

Zhu²,

Wang³

et al. 2022

Soft Comput

View full text Add to dashboard Cite

Aspect-oriented Fine-grained Opinion Extraction (AFOE) aims to extract the aspect terms, corresponding opinion terms and sentiment polarity in a target sentence. Most previous methods treat AFOE as word-level or span-level task, which ignore the complementarity of these two tasks. To integrate the merits of word-level and spanlevel information, we construct an end-to-end Span-based Multi-Table Labeling (SpanMTL) framework. SpanMTL combines word-based and span-based table labeling to tackle AFOE task. Specifically, in the proposed model, we use two separate BiLSTMs to encode the information of aspect and opinion terms into a word-based 2D representation table. Based on the table, we construct span-based table with CNN by associating the word-pair representations. At last, we integrate the table label distributions of word-and span-based table labeling to generate a multi-table labeling. The proposed method improves the performances of OPE and OTE tasks by introducing span information especially on the data with lots of spans. We have conducted various experiments on AFOE datasets to validate our Article Title method. The experimental results show that our method outperforms other baselines when the sentences having lots of span information.

show abstract

Study on a simple summative attention mechanism used in BERT model for topic classification and sentiment analysis of car reviews

HUANG,

HUANG

2023

International Conference on Electronic Materials and Information Engineering (EMIE 2023)

View full text Add to dashboard Cite

First Target and Opinion then Polarity: Enhancing Target-opinion Correlation for Aspect Sentiment Triplet Extraction

Cited by 7 publications

References 29 publications

Explicit Interaction Network for Aspect Sentiment Triplet Extraction

Explicit Interaction Network for Aspect Sentiment Triplet Extraction

SpanMTL: a span-based multi-table labeling for aspect-oriented fine-grained opinion extraction

Study on a simple summative attention mechanism used in BERT model for topic classification and sentiment analysis of car reviews

Contact Info

Product

Resources

About