Abstract: Automatic query expansion is an effective way used to improve the efficiency of the Information Retrieval system (IR) based on the Pseudo Relevance Feedback (PRF). The first-pass retrieval contains relevant and irrelevant documents. In this study, we used different algorithms to select the terms from top retrieved documents which used for expansion query. Where proximity term frequency methods based on kernel function was used to show the semantic relationship between two words and distributional methods KLD and CHI square methods were used to show the divergence of the terms in two spaces. Then we used borda combined method. After that, we used reweighting method to calculate the new weight of the candidate terms which used to expand the query. Our result produced tangible improvement in IR system.