Opinion Mining of Online Product Reviews from Traditional LDA Topic Clusters using Feature Ontology Tree and Sentiwordnet

Abstract

Online product reviews provide data about the user"s perspective on the features that were experienced by them. Product features and corresponding opinions form a major part in analyzing the online product reviews. Extracting features from a huge number of reviews is classified into three major categories such as utilizing language rules, sequence labeling as well as the topic modeling. Latent Dirichlet Allocation (LDA) is one such topic model which clusters the document words into unsupervised learned topics using Dirichlet priors. The words so clustered are the features and opinion words in the product reviews domain. To identify appropriate product features from these clusters a hierarchical, domain independent Feature Ontology Tree (FOT) is applied to LDA clusters. The opinion bearing words of obtained product features are identified by utilizing the document indicators available from topic matrix of LDA. These indicators are useful to backtrack to the corresponding online review in which the product feature is present. The polarity of the opinion bearing word is calculated with the help of SentiWordNet. This improves the accuracy of the features using extracted LDA topic clusters and machine interpretation of polarity of opinion word is satisfactory.

Probabilistic Topic Models for Enriching Ontology from Texts

“…A comparison regarding main approaches [4,13,28,37,45,46,51] with respect to our proposed criteria is shown in Table 1 and illustrates that all existing studies use a semiautomatic approach to learn domain-specific ontologies and use several criteria to evaluate ontologies. Also, the inconsistency regarding entries is still an open challenge.…”

“…We observe that, in the extraction phase, all existing systems rely only on simple terms and do not extract complex ones leading consequently to produce syntactic inconsistency. In the classification phase, we also note that [13,46] and Posch [45] use a certain document representation method which does not consider semantic relations among different words. Gutiérrez-Batista et al [28], Asfari et al [4], and Zavitsanos et al [51] use the probabilistic document representation to recognize terms of documents and to use topic probability distribution for representations.…”

“…It consists of composing a synonym list containing the NPs that are written in different ways in order use one common NP. To do so, we rely here on the NPs list reviewed by experts and consider it "correctly generated" with no [4] Semi-automatic Criteria-based evaluation Domain-specific Unstructured data Simple term Probabilistic [45] Semi-automatic Criteria-based evaluation Domain-specific Unstructured data Simple term Basic [46] Semi-automatic Criteria-based evaluation Domain-specific Unstructured data Simple term Basic [37] Semi-automatic Criteria-based evaluation Domain-specific Unstructured data Complex term Basic [28] Semi-automatic Criteria-based evaluation Domain-specific Unstructured data Simple term Probabilistic Proposed approach Semi-automatic ALL Generic Unstructured Data Complex term Probabilistic SN Computer Science errors. Thus, we highlight that one of the characteristics of our vocabulary is that each generated term has only one meaning which is represented by a single term.…”

A Trend Analysis of Machine Learning Research with Topic Models and Mann-Kendall Test

“…In [18], proposed a method for topic identification in web documents using web design features. In [19], opinion mining performed on online product reviews with LDA topic clusters using feature ontology tree and sentiwordnet. Also, in [20] Mann-Kendall test is used to understand the nature of trends in rainfall of different regions in India.…”

A Quantification Method of Composite Impression of Products by Externalized Evaluation Words of the Appraisal Dictionary with Review Text Data

“…Htay and Lynn [8] proposed a method to extract opinion words or phrases for each feature by attaining the patterns from the review text through some specific parts of speech (POS) in order to generate summarized information that helps customers make a better selection. Santosh, Babu, Prasad, and Vivekananda [9] reported that, by applying the topic model, we can construct a domain-independent Feature Ontology Tree (FOT) in order to identify the characteristics of products. Although most of these methods focused on whether the review is positive or negative, or the main topic of the review, the impressions related to affective evaluation are not yet specifically reated.…”