Improved prediction on recommendation system by creating a new model that employs Mahout collaborative filtering with content-based filtering based on genetic algorithm methods

With the access to overwhelming information available on the internet, the common user finds it difficult to extract relevant data from the repository. To deal with an increasingly complex and time-consuming data-searching process, it is imperative to find an artificial intelligence-based solution. As a result, recommender systems with sophisticated filters that learn about users’ preferences and select the most relevant information are solicited. A recommender system plays a critical role in predicting the correct item from a pool of information. In the present study, we employed a genetic algorithm for the development of a hybrid recommendation model. We amalgamated the Mahout collaborative filtering and content-based filtering to explore an effective hybrid RS. To improve the hybrid recommendation system, we relied on the preference of the nearest neighbour of an active user rather than considering the preferences of all users. We also created the first-generation selection process by adding specific criteria to accelerate the examination process. The output of the as-proposed model indicates that its implementation has high novelty, coverage, and better performance, etc. In addition, the as-proposed approach can also help to alleviate sparsity and scalability problems through a combination of mixed methods. The results show GA’s superiority and ability to achieve more correct predictions on a large volume of data.