With the explosion of information, more and more people are embarrassed to manage information effectively. How to search and retrieve accurate information match to people's requirements has been an important issue in information management research. Although search engine can solve this problem partly, the support of manage information is still limited. To use search engine, the users should input precise keywords by themselves first and this stage might cause much confusion to users. For that reason, we need a recommendation system that can catch users' preferences to help users to obtain information more quickly and conveniently without copious process. In our research, a recommendation system is designed based on users' profile. We use ontology technology to be the core of our recommendation system, because ontology can describe the concepts and relations of individual's domain knowledge. Formal Concept Analysis (FCA) algorithm is one of the most popular methods to build ontology, and we apply it to construct our experimental system to recommend master theses to subjects. In order to evaluate our recommendation system, we developed a FCA-based system and another Scoring FCA-based system as treatments, and a Keyword-based system as a control group. We focus on both users' satisfaction on information quality and system quality of our systems. The results show that users have higher information satisfaction on Scoring FCA-based system and FCA-based system than Keyword-based system. This study contributes to research and practice in information recommendation system.