Semantic Provenance Based Trustworthy Users Classification on Book-Based Social Network using Fuzzy Decision Tree

As web-based social network allows anyone to post the content without any restriction, the trustworthiness of the content creator plays an important role before using the content. An effiective way to find the trustworthiness is, by analyzing the web resources related to the content creator. Therefore the trustworthiness is assessed using the provenance based ontological model called W7 model. Since it is a real time data, the computed trust for each reviewer using the ontological model is uncertain and vague. An appropriate way to classify such data is using the fuzzy logic with gradual trust level. As the computed trust data are feature-based and non-symbolic, the classification ambiguity need to be reduced greatly. This is achieved with the fuzzy decision tree approach, which is a fusion of fuzzy sets with decision tree. The truth of the rule is crucial in trustworthy user classification, as highly truthful rules really increase the credibility of the user in their domain. Therefore, in the proposed model, degree of truth is used as a pruning criteria that classifies the users with minimum number of fuzzy evidence or knowledge. This paper proposes a semantic provenance based gradual trust model to classify the trustworthy reviewers in a book-based social networks using fuzzy decision tree approach. Performance analysis of the proposed model in the terms of classifier accuracy, precision, recall, the number of rules generated and its time complexity are discussed. The analysis shows that the proposed learning model outperforms other classification models. This method is also applied to other data sets and the performance of the classifier is assessed.