Easily accessed publishing channels have resulted in the problem of information overload. Conventional information retrieval models, such as the vector model or the probability model, apply the lexical information contained in relevance feedback in the enhancement of document re-ranking. Improvement is possible considering the application of semantic information. Studies have been taking the approach of concept extraction and application in the dealing with this semantic matter. So far, a perfect solution remains elusive and research still has new ground to cover. As such, we have proposed and tested a strategic method to form a more understanding of this field of study. The results of formal tests show that the proposed method is more effective than the baseline ranking model.