Enhancing Movie Recommendation Systems through Demographic and Content-Based Filtering Using TF-IDF and Cosine Similarity

the need of data collection in recent years has piloted into a transformative era of information utilization. The recommendation systems represent a specialized type of information filtering mechanisms designed to improve the quality of search outcomes by suggesting items that are highly related to user queries or aligned with their historical interactions. This research uses demographic filtering for a recommending popular and well rated movies to users with similar demographic backgrounds irrespective of genre or other parameters. The methodology used in this paper computes pairwise similarity scores for all movies based on the parameter of descriptions and recommend movies to the users based on that similarity score. TF * IDF is used to calculate the relative frequency that is the importance of the words in description. It vectorizes the object to generate TDF-IDF matrix of plots of movies. Cosine similarity is applied to find similarities between each pair of movies. The above approach generates a matrix in which each column represents a movie and each column has a word from the overview vocabulary that appear in at least one document. This is practiced to decrease the value of words that occur frequently in plot description, and hence reduces their impact in calculating the final similarity score. Demographic and content-based filtering is used for recommendation in this paper, while demographic filtering provides a very fundamental method for recommendation and cannot be used practically, whereas content-based filtering is more advanced and powerful.