Abstract: This paper proposes an SPARQL ontology query based on heuristic rules.In the proposed method,LTP (Language Technology Platform) is utilized to parse a question to dependency parsing tree (DPT).Heuristic query triple extraction rules are formed according to the statistic and analysis of DPTs of question set.Query triple(s) are extracted accurately by deleting meaningless query triple(s) and recombining incomplete query triple(s) based on these rules.Query triple(s) are mapped to ontology triple(s) by means of three kinds of mapping:class mapping,instance mapping and property mapping.And then,SPARQL query is obtained.Intermediate results and answer will be presented to users when they submit a Chinese natural language question in the query interface.The experiment shows that the presented method is effective.