Recommendation System for High Average Utility Itemset using Deep Learning Techniques

Abstract

Data Mining is the technology that blends the traditional Data Analysis method with the sophisticated algorithms for processing large amount of data. Recommendation system enables the ability to use the various types of user-preference and user-requirement data to make recommendation. There many issues faced by the recommendation system among that major issue is that they need large data to effective make the recommendations. The main problems that are raised are changing the data, changing user preferences and unpredictable items. To overcome, in the proposed work High Average-Utility Mining is used. The two Deep learning techniques namely Recurrent Neural Network and fusion of Recurrent Neural Network-Convolutional Neural Network have been used. For this, dataset have been obtained from online of super market. For preprocessing, normalization is done on the actual dataset followed Long Format Data Frame is implemented to find the High Average Utility Itemset Mining. The preprocessed data is given as the input to Convolutional Neural Network which is activated by the sigmoid function. On the other hand the sigmoid Function is replaced with Recurrent Neural Network. 2-Bidirectinal LSTM is used which classifies the Predicted Products and the top k recommended products. From the above experiments, it is found that CNN-RNN fusion model gives the satisfactory results.