Fuzzy Quantification and Opinion Mining on Qualitative Data using Feature Reduction

Abstract

In this paper, we propose a generic recommender system that combines opinion mining and fuzzy quantification methods for qualitative data. The proposed system has two novel aspects. First, it employs a novel semantic orientation (SO) computation method to reduce the number of extracted features and opinion expressions. By using this new SO computation method, the proposed recommender system finds out the most related features and opinion expressions. Second, the proposed system generates short summary sentences from qualitative data using fuzzy quantification. The proposed system is evaluated using a restaurant review dataset. The results present that fuzzy quantified sentences offer brief information about the restaurant features from customers’ feedback. In addition, opinion mining extracts positive, negative, and neutral emotions from reviews.

Average utility driven data analytics on damped windows for intelligent systems with data streams

“…2 Data mining is a process of discovering important knowledge from enormous amount of data, and it is drawing interests as a major research subject. For example, the areas of data mining contain opinion mining, 3,4 granular computing-based data mining, 5,6 and web mining. 7 Association rule mining, [8][9][10] which is a branch of data mining, finds the relationship between items in a database.…”

Efficiently mining erasable stream patterns for intelligent systems over uncertain data

“…Data mining is used to obtain meaningful information needed for the system. For example, examples of data mining are opinion mining, 6,7 granular computing framework for mining relational data, 8,9 and outlier reduction in web mining. 10 Pattern mining, which is one of the data mining techniques, is a method that is used to discover meaningful data in the form of patterns.…”