In recent years, semantic search has been successfully used for information retrieval, however, it is still in the early stage, and there is not semantic search engine for minorities in China. In this paper, we propose a semantic retrieval model which is comprised of resources collection, semantic annotation, query analysis and results ranking. For semantic annotation, we use domain ontology and two bilingual dictionaries to extract keywords for annotation. For query analysis, we present a method which combines lexical relationship and semantic relationship to analyse user's query. And for results ranking, we propose a modulative method that ranks results based on how predictable a result might be for users, which is a blend of semantic and information-theoretic techniques. The preliminary experimental results show the capability of the proposed model to boost the precision and recall rates of webpage searching.