Constrained LDA for Grouping Product Features in Opinion Mining

Zhai, Zhongwu; Liu, Bing; Xu, Hua; Jia, Peifa

doi:10.1007/978-3-642-20841-6_37

Cited by 105 publications

(53 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, Zhai et al [28] proposed another semi-supervised method known as constrained-LDA [24] that does not need any pre-labeled data as initial seed. Constrained-LDA is an extension of LDA method with the ability to consider existing knowledge in the form of must-link and cannot-link constraints.…”

Section: Semi-supervised Approachesmentioning

confidence: 99%

See 1 more Smart Citation

Literature Review on Feature Identification in Sentiment Analysis

Hussain¹,

Sattar²,

Afzal³

2015

IJCA

View full text Add to dashboard Cite

Now-a-days the volume of opinions about products, issues, events, and politics etc. on different social, e-commerce and review sites grows very rapidly. From both opinion holder and opinion target point of view, it is very difficult and time consuming task to analyze all the reviews from this massive amount of data on the web. So, there is a need of efficient method that automatically extracts the opinions and relevant features of the opinion target from the reviews and finally generates the feature wise summary. Sometimes people may use different words to express same feature, this may produce a misperception in the results during feature wise summary generation. To avoid this, we need to categorize similar features for precise classification of opinions based on these feature groups. Therefore, our study is targeting the most important tasks of feature based sentiment analysis that are feature extraction and feature categorization. This paper is about to cover the currently available techniques in these two areas. We have also focused on least addressed area in this domain giving an opportunity to researchers for future work. General TermsPattern Recognition

show abstract

Section: Semi-supervised Approachesmentioning

confidence: 99%

“…This observation allows us to extract many "cannot-link" constraints automatically. Secondly for "must-links", it can be observed that two noun phrases that shared one or more words are likely to fall into the same topic, e.g., "battery life", "battery consumption" and "battery power" [28].…”

Section: Semi-supervised Approachesmentioning

confidence: 99%

Literature Review on Feature Identification in Sentiment Analysis

Hussain¹,

Sattar²,

Afzal³

2015

IJCA

View full text Add to dashboard Cite

show abstract

“…Thus the sellers can be ranked efficiently as if ranking is done manually. The correlation between natural (manual) clustering and CommTrust clustering [13] was checked using Kendall"s τ measure. The correlation was found positive i.e.…”

Section: Dmentioning

confidence: 99%

Literature Analysis on Reputation Models for Feedback in E-commerce

Thomas¹,

Joseph²

2015

International Journal of Advanced Research in Computer and Comm

View full text Add to dashboard Cite

Trust and reputation are decisive factors in internet service provision. The concept is to let the parties rate each other. Reputation models used in e-commerce have the main intention to guide the customers or buyers in a proper way so that they can choose between the different available trustworthy sellers. The models are meant to project the different sellers as per their services and quality. Models are a real need because most of the e-commerce systems have different qualities. But models differ between each other by the methods they opt to calculate the trustworthiness measure for different sellers. Based on such calculations the ranking of the sellers will also vary. Feedback mining method of trust calculation outperforms all other methods with the idea that users are free to express themselves in textual feedback comments. The method addresses the issue of positive bias which has become a major concern nowadays in the field of e-commerce.

show abstract

“…• Constraint-LDA: Constraint-LDA (Zhai et al, 2011b) is a state-of-the-art LDA-based method that incorporates must-link and cannot-link constraints for this task. We set the damping factor λ = 0.3 and relaxation factor η = 0.9, as suggested in the original reference.…”

Section: Comparison To Unsupervised Baselinesmentioning

confidence: 99%

Clustering Aspect-related Phrases by Leveraging Sentiment Distribution Consistency

Zhao¹,

Huang²,

Chen³

et al. 2014

Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)

View full text Add to dashboard Cite

Clustering aspect-related phrases in terms of product's property is a precursor process to aspect-level sentiment analysis which is a central task in sentiment analysis. Most of existing methods for addressing this problem are context-based models which assume that domain synonymous phrases share similar co-occurrence contexts. In this paper, we explore a novel idea, sentiment distribution consistency, which states that different phrases (e.g. "price", "money", "worth", and "cost") of the same aspect tend to have consistent sentiment distribution. Through formalizing sentiment distribution consistency as soft constraint, we propose a novel unsupervised model in the framework of Posterior Regularization (PR) to cluster aspectrelated phrases. Experiments demonstrate that our approach outperforms baselines remarkably.

show abstract

Constrained LDA for Grouping Product Features in Opinion Mining

Cited by 105 publications

References 17 publications

Literature Review on Feature Identification in Sentiment Analysis

Literature Review on Feature Identification in Sentiment Analysis

Literature Analysis on Reputation Models for Feedback in E-commerce

Clustering Aspect-related Phrases by Leveraging Sentiment Distribution Consistency

Contact Info

Product

Resources

About