A major challenge for creating personalized diet and activity applications is to capture static, semi-static and dynamic information about a person in a user-friendly way. Sharing and reusing information between heterogeneous sources like social networking applications, personal health records, specialized applications for diet and exercise monitoring, and personal devices with attached sensors can achieve a better understanding of the user. Gathering distributed user information from heterogeneous sources and making sense of it to enable user model interoperability entails handling the semantic heterogeneity of the user models. In this paper, we enhance the process of concept alignment to automatically determine semantic mapping relations to enable interoperability between heterogeneous health and fitting applications. We add an internal structure similarity measure to increase the quality of generated mappings of our previous work. We show that the addition of an internal structure analysis of source data in the process of concept alignment improves the efficiency and effectiveness of measuring results. Constrain and data type verification done in the internal structure analysis proved to be useful when dealing with common conflicts between concepts.