By integrating data of sci-tech novelty retrieaval and internet data, the paper makes a precise analysis of the potential demands, interest, interest, tendency,preferences and habits of users of novelty retrieval. From the perspective of big data technology, it analyzes the feasibility of designing targeted and deep-level services for mass user data, puts forward the framework and method of conducting deep-level and precise service recommendation based on users of sci-tech novelty retrieval, and studies the opportunity to provide these users with deep-level and precise services.