Amazing technological breakthrough possible @S-Logix

Office Address

  • #5, First Floor, 4th Street Dr. Subbarayan Nagar Kodambakkam, Chennai-600 024 Landmark : Samiyar Madam
  • +91- 81240 01111

Social List

Aspect-Based Opinion Mining From Customer Reviews

Research Area:  Machine Learning


Peoples opinions and experience are important sources of information in our everyday life.In the modern digital age, text is the main method of communicating information on the Internet.Typically, people ask their family and friends about which smart phone, restaurant or doctor they would recommend. Opinionated information abounds on the Internet, including productor service reviews, blogs, feedback forms, Facebook likes and more. It is common practice forpeople to share their opinions about the products or services they have used. Customer reviewsare publicly available on most online trading websites to assist others with their purchasedecisions.This abundance of information creates the need for a mechanism to extract and summariseuseful data to help make informed decisions. With rapid development in the areas of informationretrieval, data mining and natural language processing, the opportunities for summarising in-formation automatically have greatly increased. The research in this thesis investigates new ap-proaches to automated analysis, aggregation, and extraction of opinions and aspects of customerreviews from text using data mining and natural language processing techniques. It focuses onaspect-based opinion mining from customer reviews. It discusses the characteristics of customerreviews and describes different methods to extract aspects and corresponding opinions.Data mining, natural language processing and statistical approaches are utilised based ondictionaries, grammatical analysis and semantic understanding of text. In particular, association rule mining is used to discover interesting relations among different entities within the datasetbased on the most highlighted rules. Furthermore, dependency relations are used based on the grammatical representation of sentence structure as a set of relationships among entities. Con-ditional Random Field is used to encode dependencies between different entities of a sequence via probabilistic relations.

Name of the Researcher:  Zhen, Hai

Name of the Supervisor(s):  ProfessorYuefeng Li, Dr Jinglan Zhang and Dr Yue Xu

Year of Completion:  2016

University:  Queensland University of Technology

Thesis Link:   Home Page Url