Detection of Opinion Spam with Character n-grams

Fusilier, Donato Hernández; Montes-y-Gómez, Manuel; Rosso, Paolo; Guzmán-Cabrera, Rafael

doi:10.1007/978-3-319-18117-2_21

Cited by 44 publications

(45 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Different linguistic features can be considered; the simplest ones are either individual words or n ‐grams, (i.e., groups of n words), which can be either weighted or unweighted . For a complete overview on the multiple kinds of linguistic features that have been proposed in the literature, refer to Crawford et al The majority of purely content‐based approaches are based on supervised learning techniques …”

Section: Approaches To Credibility Assessmentmentioning

confidence: 99%

“…On the same dataset, Banerjee and Chua use logistic regression (LR) to build a classification model by considering the readability of a review (i.e., its complexity and reading difficulty), the distribution of POS tags, and the review writing style, i.e., positive cues, perceptual words, and usage of future tense. Fusilier et al use both character n ‐grams and word n ‐grams obtaining the best results with character n ‐grams with values for n of 4 and 5, respectively, by using NB and an SVM classifier on the Ott et al’s dataset…”

Section: Approaches To Credibility Assessmentmentioning

confidence: 99%

“…30,33 For a complete overview on the multiple kinds of linguistic features that have been proposed in the literature, refer to Crawford et al 32 The majority of purely content-based approaches are based on supervised learning techniques. [34][35][36][37] Ott et al 34 use three groups of features to train Nave Bayes (NB) and support vector machine (SVM) classification algorithms. For each review, the authors define linguistic features that are based on the frequencies of each parts-of-speech (POS) tag, 38 and they test differences in POS distribution between fake and nonfake reviews based on the genre of reviews.…”

Section: Content-based Approachesmentioning

confidence: 99%

“…On the same dataset, Banerjee and Chua 36 use logistic regression (LR) to build a classification model by considering the readability of a review (i.e., its complexity and reading difficulty), the distribution of POS tags, and the review writing style, i.e., positive cues, perceptual words, and usage of future tense. Fusilier et al 37 use both character ngrams and word n-grams obtaining the best results with character n-grams with values for n of 4 and 5, respectively, by using NB and an SVM classifier on the Ott et al's dataset. 34 It is worth to be noticed that Mukherjee et al 40,41 and Sun et al 42 observed that using the AMT for producing fake reviews causes inaccurate results in review spam detection, because fake reviews generated ad hoc (e.g., by means of AMT) are not meaningful in spam detection techniques when compared with real reviews collected from opinion-sharing sites.…”

Section: Content-based Approachesmentioning

confidence: 99%

See 3 more Smart Citations

Credibility in social media: opinions, news, and health information—a survey

Viviani

Pasi

2017

WIREs Data Min & Knowl

154

View full text Add to dashboard Cite

In the Social Web scenario, where large amounts of User Generated Content diffuse through Social Media, the risk of running into misinformation is not negligible. For this reason, assessing and mining the credibility of both sources of information and information itself constitute nowadays a fundamental issue. Credibility, also referred as believability, is a quality perceived by individuals, who are not always able to discern with their cognitive capacities genuine information from the fake one. For this reason, in the recent years several approaches have been proposed to automatically assess credibility in Social Media. Most of them are based on data‐driven models, i.e., they employ machine‐learning techniques to identify misinformation, but recently also model‐driven approaches are emerging, as well as graph‐based approaches focusing on credibility propagation. Since multiple social applications have been developed for different aims and in different contexts, several solutions have been considered to address the issue of credibility assessment in Social Media. Three of the main tasks facing this issue and considered in this article concern: (1) the detection of opinion spam in review sites, (2) the detection of fake news and spam in microblogging, and (3) the credibility assessment of online health information. Despite the high number of interesting solutions proposed in the literature to tackle the above three tasks, some issues remain unsolved; they mainly concern both the absence of predefined benchmarks and gold standard datasets, and the difficulty of collecting and mining large amount of data, which has not yet received the attention it deserves. WIREs Data Mining Knowl Discov 2017, 7:e1209. doi: 10.1002/widm.1209 This article is categorized under: Algorithmic Development > Web Mining Application Areas > Science and Technology Technologies > Machine Learning

show abstract

Section: Approaches To Credibility Assessmentmentioning

confidence: 99%

Section: Approaches To Credibility Assessmentmentioning

confidence: 99%

Section: Content-based Approachesmentioning

confidence: 99%

Section: Content-based Approachesmentioning

confidence: 99%

See 2 more Smart Citations

Credibility in social media: opinions, news, and health information—a survey

Viviani

Pasi

2017

WIREs Data Min & Knowl

154

View full text Add to dashboard Cite

show abstract

“…In the relatively new field of fake review detection, especially in real‐life Web sites, much of the previous work focuses on techniques based on review text analysis and behavioral/meta‐data analysis by applying supervised, semi‐supervised, and unsupervised machine learning techniques. Regarding supervised approaches that take into account the text of the reviews, some works analyze the actual content of the reviews, modeling it with n ‐gram feature sets . Other works are based on the identification of duplicate reviews, considered as fake reviews .…”

Section: Related Workmentioning

confidence: 99%

Quantifier Guided Aggregation for the Veracity Assessment of Online Reviews

Viviani

Pasi

2016

Int. J. Intell. Syst.

View full text Add to dashboard Cite

The Social Web is characterized by a massive diffusion of unfiltered content, directly generated by users via the spread of different social media platforms. In this context, a challenging issue is to assess the veracity of the information generated within the sites of online reviews. To address this issue, a common practice in the literature is to select and analyze some veracity features associated with users and their reviews, by mostly applying machine learning techniques, to provide a classification in genuine and deceptive reviews. In this paper, we do not focus on the feature selection and user behavior analysis issues, but we concentrate on the aggregation process with respect to each single veracity feature. In most of the approaches based on machine learning techniques, the contribution of each feature in the classification process is not measurable by the user. For this reason, we propose a multicriteria decision making approach based both on the assessment of multiple criteria and the use of aggregation operators with the aim of obtaining a veracity score associated with each review. Based on this score, it is possible to detect fake reviews. The proposed model is evaluated on a Yelp data set by applying different aggregation schemes, and it is compared with well‐known supervised machine learning techniques.

show abstract