Published April 30, 2020 | Version v1
Journal article Open

Personalized Recommendation System based Association Rule Mining and Sentiment Classification

  • 1. P.hD Scholar, Department of Computer Science, VELS Institute of Science, Technology and Advanced Studies, Chennai.
  • 2. M.Phil Computer Science, Manonmanium Sundaranar University, Tirunelveli.
  • 1. Publisher


In this emerging global economy, e-commerce is an inevitable part of the business strategy. Moreover, the business world comprises the upcoming entrepreneurs who are unaware of the current trends in marketing. Therefore, a recommendation system is very essential for them. In this paper, a fully automated recommendation system for the upcoming entrepreneurs to become successful in their business is proposed. The system works in three stages. In the first stage, the most transacted product is identified using association rule mining FP growth algorithm. This helps in extracting useful information from the previous transacted data by mining the entire set of frequent patterns. The second stage identifies the most customer preferred company based on review analysis. The multilevel clustering process with the generalization of data review is implemented to achieve an accurate review of the product. It rectifies the problems of shilling attack and gray sheep users commonly seen in single level K-means algorithm by refining the collected data. In the third stage, the reviews are sorted using a polarity shift sentiment classification algorithm. It helps in sort positive and negative reviews thereby rating a company. The top rated company would give the best product. Thus, the best product can be identified. From the experimental analysis, it is understood that the proposed system outperforms the existing recommendation methods. Moreover, this automated system helps the user to get the most accurate result within time. Hence, it would be very beneficial to the upcoming businessmen for flourishing their business in this increasing economic world.



Files (594.7 kB)

Name Size Download all
594.7 kB Preview Download

Additional details

Related works

Is cited by
Journal article: 2249-8958 (ISSN)


Retrieval Number