SVR-based outlier detection and its application to hotel ranking

Abstract

With the rapid advance in information technology, more and more information exchange platforms appear. People can freely exchange information on these platforms. However, not all information is reliable. To make correct decisions, it is necessary to detect and remove unreliable information. The main purpose of this study is to improve the reliability of hotel ranking by detecting and deleting outlier on-line reviews. For this purpose, we design a support vector regression (SVR) based outlier detector using existing on-line reviews. Intuitively, normal reviews are regular, and can be correctly labeled by the SVR detector. Outlier reviews, on the other hand, are usually not regular, and cannot be correctly labeled. Thus, a well-designed SVR-detector can help us to delete outlier reviews automatically. Results obtained in this study are useful not only for hotel ranking. In principle it can be good for recommendation of any services.

Computational Intelligence in the hospitality industry: A systematic literature review and a prospect of challenges

“…A plethora of ML techniques have been applied in this area, but the 34 per cent of the analyzed research works on this topic expose the use of SVM algorithms. In [130][131][132]134,135,138,141,142] and [147] SVMs are used to extract sentiment data from reviews, blogs and various platforms, and in [74,77,85,90,95,133,137,140] and [98] SVMs are used along with N-Gram kernels, various naive Bayes techniques, ANNs, association rules, maximum entropy classifiers, C4.5 decision trees and random forests, JRIP and SVR.…”

A Survey on Machine Learning‐Based Mobile Big Data Analysis: Challenges and Applications

“…Manual removal of error data is difficult and impossible to accomplish in MBD due to the massive volume. Common data cleaning methods can alleviate the dirty data problem to some extent by training support vector regression (SVR) classifiers [73], multiple linear regression models [74], autoencoder [75], Bayesian methods [76][77][78], unsupervised methods [79], or information-theoretic models [79].…”