The quality assessment tools take an important role in software engineering, but the applicability of static analysis tools to large code bases is still questioned. Offloading semantic information to a separate database in order to free up RAM, as well as an organization of long-term storage, become important tasks. In this paper, the focus is put on one of the approaches to storing such data based on NoSQL databases. Such a system may compactly and effectively store complex semantic objects with weak linkage. Entity schemas are described, the metadata serving as a link between different analysis types is conceptualized. The advantage of the model is that it allows compact storing of nested objects, flexibility due to the ability to use distributed technology stack. The implementation was successfully tested on Linux kernel. The comparison of the approach on different size projects was performed and presented in the paper. The results are two-fold: at one hand, small project verification slows down due to ove