Abstract: Currently, there is an enormous sea of big data in engineering technology. It is of great importance to integrate the existing digital resources and to assist the experts and scholars to find the current hot spots of research in various fields with knowledge discovery technology. This paper proposes a bayesian-statistical-inference-based keyword association algorithm by measuring the direct relationship and latent semantic relationship between keywords. The proposed method is applied for solving the problemsof user behavior information shortage and co-occurrence sparsity.

Key words: Knowledge discovery, Keyword recommendation, Latent semantic, Co-occurrence sparsity, Data mining