Recommender System The rapid expansion of Internet contents’ scalability has impacted on the growing number of accessed data and dispersed information. Simultaneously Internet has become a solitary source in providing useful information in order to enrich electronic documents when accessed by users. However for the most of users, defining specific term in searching information related to electronic document is not a simple task to be accomplished. Moreover with the large number of data dispersed in Internet, the process to obtain the related term is very time consuming. Therefore, an effective and rapid method is very important to accurately extract electronic documents information to define a specific term needed by users. Approaching the solution for this  issue, this project proposes a system that has a capability to extract semantic information from electronic document. This\ system performs a method to analyze document’s information and  correlate it to the term required by the users. Since electronic documents could be large, the system approaches the clustering  method to distribute related documents information. Some rules are applies to classify and sort the documents into a particular cluster, based on its correlation. Features of this system include the pre-process method using Part-of-Speech Tagger as the equipment for word recognition, semantic word tool with Natural Language Processing (NLP) as the abstract extractor; and machine learning method as the word constructing tool.