- مبلغ: ۸۶,۰۰۰ تومان
- مبلغ: ۹۱,۰۰۰ تومان
The Internet is host to many sites that collect vast amounts of opinions about products and services. These opinions are expressed in written language, and this paper presents a method for modeling the aspects of overall customer satisfaction from free-form written opinions. Written opinions constitute unstructured input data, which are first transformed into semi-structured data using an existing method for aspect-level sentiment analysis. Next, the overall customer satisfaction is modeled using a Bayesian approach based on the individual aspect rating of each review. This probabilistic method enables the discovery of the relative importance of each aspect for every unique product or service. Empirical experiments on a data set of online reviews of California State Parks, obtained from TripAdvisor, show the effectiveness of the proposed framework as applied to the aspect-level sentiment analysis and modeling of customer satisfaction. The accuracy in terms of finding the significant aspects is 88.3%. The average R2 values for predicted overall customer satisfaction using the model range from 0.892 to 0.999.
The major contribution of this paper is the linking of aspect identification and semantic classification methods to explain and predict overall customer satisfaction. First, a method is proposed by which unstructured user generated text data is transformed into ready-to-analyze data without the need to determine aspects a priori. Second, a Bayesian model is proposed that allows prediction of individual aspect ratings, and further enables discovery of the relative im475 portance of each aspect from each contributor’s perspective. Consequently, the method also allows for prediction of overall customer satisfaction. The model presented in this paper has low dimensionality that can be scaled to analyze very large data sets in an automated fashion. Results of the Bayesian method are reproducible. Thorough testing illustrates that the methods presented in this paper are effective in discovering, explaining, and predicting the most important aspects driving overall customer satisfaction.