The major elements or a concept-based retieval model UE a concepl eqlomtion algorithm and a howledge base smctured as a semantic network, n u s sbdy pfomed two expiments mmpmng d m concept ehplomtion algoriIllms and rhen four !mwledge bases h ordcr to h d d~e most eIfectve concept-based rehievd mo&l Experiments are divided into hvo pups according lo tl~e size of a texc collecljon Fach expedmcnt is suwiv~ded Into algonlhmcompfag expimlt and knowledge basecomparing expedment. Jn he algorih-compadng eqeiment, thc pfo~mmces 01 uhe qualtial bnb algonihq the p d e l bnb dgorillun, andthc HopF~eld net algonilun were evaluated and compared with the P-nonn re11wa1 rncdel which is an extended the Bwlean rehievd model. Jn the knowledge base-compmg exp(lvne11C he peiiormances of v;uiws knowledge bares were mmwd using a selecled algonh which had been identkd as the most efihve in the first expmeal. The fow types of )nowledge k s developed for flus expeimcnt xc documeul-based ihesauius-W, iluqare4 and synonym lxwessd