Summarizing Opinions: Aspect Extraction Meets Sentiment Prediction and They Are Both Weakly Supervised

Angelidis, Stefanos; Lapata, Mirella

doi:10.18653/v1/d18-1403

Cited by 155 publications

(191 citation statements)

References 33 publications

Supporting

Mentioning

187

Contrasting

Unclassified

Order By: Relevance

“…• MATE-*: The MATE model of Angelidis and Lapata (2018) with various configurations: initialization of the aspect embeddings A k using the unweighted/weighted average of seed word embeddings and an extra multitask training objective (MT). 8…”

Section: Experimental Settingsmentioning

confidence: 99%

“…The average number of training, validation, and test segments across domains is around 1 million, 700, and 700 segments, respectively. Segment statistics per domain are reported in the supplementary material of Angelidis and Lapata (2018).…”

Section: A1 Datasetsmentioning

confidence: 99%

“…For a fair comparison, for the product reviews we use the 200-dimensional word2vec embeddings provided by Angelidis and Lapata (2018) and the base uncased BERT model. 10 For the restaurant reviews, we use the 300-dimensional multilingual word2vec embeddings provided by Ruder et al (2016) and the multilingual cased BERT model.…”

Section: A2 Implementation Detailsmentioning

confidence: 99%

“…Given the vast number of online reviews, both sellers and customers would benefit from automatic methods for detecting fine-grained segments that discuss particular aspects of interest. Fine-grained aspect detection is also a key task in downstream applications such as aspect-based sentiment analysis and multi-document summarization (Hu and Liu, 2004;Liu, 2012;Pontiki et al, 2016;Angelidis and Lapata, 2018).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Leveraging Just a Few Keywords for Fine-Grained Aspect Detection Through Weakly Supervised Co-Training

Karamanolakis

Hsu

Gravano

2019

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen

View full text Add to dashboard Cite

User-generated reviews can be decomposed into fine-grained segments (e.g., sentences, clauses), each evaluating a different aspect of the principal entity (e.g., price, quality, appearance). Automatically detecting these aspects can be useful for both users and downstream opinion mining applications. Current supervised approaches for learning aspect classifiers require many fine-grained aspect labels, which are labor-intensive to obtain. And, unfortunately, unsupervised topic models often fail to capture the aspects of interest. In this work, we consider weakly supervised approaches for training aspect classifiers that only require the user to provide a small set of seed words (i.e., weakly positive indicators) for the aspects of interest. First, we show that current weakly supervised approaches do not effectively leverage the predictive power of seed words for aspect detection. Next, we propose a student-teacher approach that effectively leverages seed words in a bag-of-words classifier (teacher); in turn, we use the teacher to train a second model (student) that is potentially more powerful (e.g., a neural network that uses pre-trained word embeddings). Finally, we show that iterative co-training can be used to cope with noisy seed words, leading to both improved teacher and student models. Our proposed approach consistently outperforms previous weakly supervised approaches (by 14.1 absolute F1 points on average) in six different domains of product reviews and six multilingual datasets of restaurant reviews.

show abstract

Section: Experimental Settingsmentioning

confidence: 99%

Section: A1 Datasetsmentioning

confidence: 99%

Section: A2 Implementation Detailsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Leveraging Just a Few Keywords for Fine-Grained Aspect Detection Through Weakly Supervised Co-Training

Karamanolakis

Hsu

Gravano

2019

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen

View full text Add to dashboard Cite

show abstract

“…Prior to the deep learning era, these information were used as effective categorical features Tan et al, 2011;Gao et al, 2013;Park et al, 2015) for the machine learning model. Recent work has used them to improve the overall performance (Chen et al, 2016;Dong et al, 2017), interpretability (Amplayo et al, 2018a;Angelidis and Lapata, 2018), and personalization (Ficler and Goldberg, 2017) of neural network models in different tasks such as sentiment classification (Tang et al, 2015), review summarization (Yang et al, 2018a), and text generation (Dong et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Rethinking Attribute Representation and Injection for Sentiment Classification

Amplayo¹

2019

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen

View full text Add to dashboard Cite

Text attributes, such as user and product information in product reviews, have been used to improve the performance of sentiment classification models. The de facto standard method is to incorporate them as additional biases in the attention mechanism, and more performance gains are achieved by extending the model architecture. In this paper, we show that the above method is the least effective way to represent and inject attributes. To demonstrate this hypothesis, unlike previous models with complicated architectures, we limit our base model to a simple BiLSTM with attention classifier, and instead focus on how and where the attributes should be incorporated in the model. We propose to represent attributes as chunk-wise importance weight matrices and consider four locations in the model (i.e., embedding, encoding, attention, classifier) to inject attributes. Experiments show that our proposed method achieves significant improvements over the standard approach and that attention mechanism is the worst location to inject attributes, contradicting prior work. We also outperform the state-of-the-art despite our use of a simple base model. Finally, we show that these representations transfer well to other tasks 1 .

show abstract

Semi-supervised Extractive Question Summarization Using Question-Answer Pairs

Machida

Ishigaki

Kobayashi³

et al. 2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Neural extractive summarization methods often require much labeled training data, for which headlines or lead summaries of news articles can sometimes be used. Such directly useful summaries are not always available, however, especially for user-generated content, such as questions posted on community question answering services. In this paper, we address an extractive summarization (i.e., headline extraction) task for such questions as a case study and consider how to alleviate the problem by using question-answer pairs, instead of missing-headline pairs. To this end, we propose a framework to examine how to use such unlabeled paired data from the viewpoint of training methods. Experimental results show that multi-task training performs well with undersampling and distant supervision.

show abstract

Summarizing Opinions: Aspect Extraction Meets Sentiment Prediction and They Are Both Weakly Supervised

Cited by 155 publications

References 33 publications

Leveraging Just a Few Keywords for Fine-Grained Aspect Detection Through Weakly Supervised Co-Training

Leveraging Just a Few Keywords for Fine-Grained Aspect Detection Through Weakly Supervised Co-Training

Rethinking Attribute Representation and Injection for Sentiment Classification

Semi-supervised Extractive Question Summarization Using Question-Answer Pairs

Contact Info

Product

Resources

About