With the explosive growth of data, it has become very difficult for a person to process the data and find the right information from it. So, to discover the right information from the colossal amount of data that is available online, we need information filtering systems. Recommendation systems (RS) help users find the most interesting information among the options that are available. Ratings given by the users play a vital role in determining the purposes of recommendations. Earlier, researchers used a user’s rating history to predict unknown ratings, but recently a user’s review has gained a lot of attention as it contains a lot of relevant information about a user’s decision. The proposed system makes an attempt to deal with the problem of uncertainty in the rating histories by using textual reviews. Two datasets are used to experimentally analyze the proposed framework. In this approach, clustering techniques are used with natural language processing (NLP) for prediction. It also compares how different algorithms, such as K-mean, spectral, and hierarchical clustering algorithms, produce a varied outcome and concludes which method is appropriate for the given recommendation scenarios. We also validate how the proposed method outperforms the non-clustering-based methods.
