With the development of Web 2 and social networks, customers and users can share their opinions about different products
They leave. These ideas can be used as a valuable resource to determine the position of the product and its success in marketing. Extracting the rep More
With the development of Web 2 and social networks, customers and users can share their opinions about different products
They leave. These ideas can be used as a valuable resource to determine the position of the product and its success in marketing. Extracting the reported shortcomings from the large volume of comments generated by users is one of the major problems in this field of research. By comparing the products of different manufacturers, customers and consumers express the strengths and weaknesses of the products in the form of positive and negative comments. Classification of comments based on positive and negative sensory words in the text does not lead to accurate results without reference to documents containing a defect report. Because defects are not reported solely in negative comments. It is possible for a customer to feel positive about a product and still report a defect in their opinion. Therefore, another challenge of this research field is the correct and accurate classification of opinions. To solve these problems and challenges, this article provides an effective and efficient way to extract comments containing product defect reports from users' online comments. For this purpose, stochastic forest classifiers were used to identify the defect report and the unattended thematic modeling technique used the Dirichlet hidden allocation to provide a summary of the defect report. Data from the Amazon website has been used to analyze and evaluate the proposed method. The results showed that random forest has an acceptable performance for defect reporting even with a small number of educational data. Results and outputs extracted from documents containing the defect report, including a summary of the defect report to facilitate manufacturers' decision making, finding patterns of the defect report in the text automatically, and discovering the aspects of the product that reported the most defects Related to themDemonstrates the ability of Dirichlet's latent allocation method.
Manuscript profile
