Detecting Fake Reviews Based on Review-Rating Consistency and Multi-dimensional Time Series

Fang, Youli; Wang, Hong; Di, Ruitong; Wang, Lutong; Jin, Li

doi:10.1007/978-3-030-05234-8_14

Cited by 2 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The shortcomings of this algorithm lied on the discriminant threshold of "Click farm fraud cannot be automatically adjusted to identify suspected fraud." Fang et al [13] brought forward a method for identifying deceptive reviews that integrates scoring evaluation consistency and multidimensional time series. Moreover, they constructed a model of deceptive review detection based on multi-dimensional time series.…”

Section: State Of the Artmentioning

confidence: 99%

Identification of Deceptive Reviews by Sentimental Analysis and Characteristics of Reviewers

Gao¹,

Li²,

Zhu³

et al. 2019

JESTR

View full text Add to dashboard Cite

The purchase decisions of customers are influenced by the relevant reviews made by customers. Deceptive reviews are confusing and hard to detect. The existing identification method of deceptive reviews is based on the traditional algorithm of machine learning. However, the methods used to identify deceptive reviews must be improved. In order to improve the accuracy of the identification of deceptive reviews, a novel method integrating sentimental analysis and the characteristics of reviewers was proposed in this study. On the basis of the analysis of the emotional characteristics of the review texts and the behavioural characteristics of the reviewers, a method of deceptive review identification was established. The proposed method analyzed the intensity of emotions, the text similarity, the largest daily publishing comment index, and the extreme rating index on the basis of the feature-weighted model. This model verified the effectiveness of the proposed method. Results show that a direct correlation exists between the unreliability score of users and deceptive review identification. If the score exceeds 0.78, then the reviewer is deemed to be a deceptive reviewer, and the reviews made are deceptive reviews. The proposed method provides a good prospect to identify deceptive reviews.

show abstract

Section: State Of the Artmentioning

confidence: 99%