Sentiment Analysis Meets Semantic Analysis: Constructing Insight Knowledge Bases Completed Research Paper

摘要

Numerous Web 2.0 applications collect user opinions, and other user-generated content in the form of product reviews, discussion boards, and blogs, which are often captured as unstructured data. Text mining techniques are important for analyzing users' opinions (sentiment analysis) and identifying topics of interest (semantic analysis). However, little work has been carried out that combines semantics with user's sentiments. This research proposes a Sentiment-Semantic Framework that incorporates results from both semantic and sentiment analysis to construct a knowledge base of insights gained from integrating the information extracted from each type of analysis. To evaluate the framework, a prototype is developed and applied to two different domains (e-commerce and politics) and the resulting insight knowledge bases constructed.