Personalized Research Paper Recommendation System using Keyword Extraction Based on UserProfile

In this paper, we proposed PRPRS (Personalized Research Paper Recommendation System) that designed expansively and implemented a UserProfile-based algorithm for extracting keyword by keyword extraction and keyword inference. If the papers don't have keyword section, we consider the title and text as an argument of keyword and execute the algorithm. Then, we create the possible combination from each word of title. We extract the combinations presented in the main text among the longest word combinations which include the same words. If the number of extracted combinations is more than the standard number, we used that combination as keyword. Otherwise, we refer the main text and extract combination as much as standard in order of high Term-Frequency. Whenever collected research papers by topic are selected, a renewal of UserProfile increases the frequency of each Domain, Topic and keyword. Each ratio of occurrence is recalculated and reflected on UserProfile. PRPRS calculates the similarity between given topic and collected papers by using Cosine Similarity which is used to recommend initial paper for each topic in Information retrieval. We measured satisfaction and accuracy for each system-recommended paper to test and evaluated performances of the suggested system. Finally PRPRS represents high level of satisfaction and accuracy.