The technologies of predicting query performance try to predict the performances of queries before actual time-consuming information retrievals

and adjust the original queries based on the predicting results to improve the performances without modifying the information needs.For avoiding the terms-independence assumption in traditional query performance predicting models

this paper proposes a query performance predicting model using semantic charts of queries which describe the semantic relations between query terms.The experimental results show that the precisions of semantic charts weighted query performance predicting models are higher than traditional models significantly

and the relativities between predicting performances and actual performances are improved about 46.32% at most.