Mohammad Reza Gholamian - School of industrial engineering, Iran University of Science and Technology
Zahra Rafat Panah - School of industrial engineering, Iran University of Science and Technology

The web has become a rich source of customer opinions and complaints about products and services that are provided by different businesses in different fields. Analyzing these opinions can help businesses in making decisions and accessing beneficial information. The aim of this article is to focus on online customer opinions in the field of organizations offering services to people, and pursues two aims. The first aim is proposing a semi supervised approach for extracting customer experiences over a number of touchpoints through finding and exploiting varying opinion segment patterns that are used by customers to describe their opinions and finally visualizing the results as a hierarchical graph. The second aim is providing a recommender system that can specify services which need to be improved or the ones which have had a good performance. The intention of this approach is to enable customers and organization managers to easily view the detailed extracted information as well as viewing general results. Evaluations of results show that the proposed technique in this article outweighs the results obtained by machine learning methods.

Complaint Management, Data Mining, Opinion Mining, Recommender Systems, Natural Language Processing, Text Mining