Many　researchers　make　use　of　WordNet　to　measure　semantic　distance,　but　rarely　describe　and,　further　more,　analyze　an　algorithm　it　which　is　it?　specialty　to　find　paths,　called　semantic　relations　in　this　paper,　between　two　concepts　it?　W＇ordNet.　Searching　in　all　semantic　relations　is　an　important　step　in　measuring　semantic　similarity.　In　this　paper,　we　propose　two　algorithms,　HS　(Hierarchy,　Spread)　and　BDOS　(Bi-Direction　One　Step),　to　search　the　relation　with　shortest　semantic　distance　it?　WordNet.　HS　takes　Hyponym　and　Hypernym　into　consideration　at　first,　while　BDOS　searches　semantic　relations　from　two　start　concepts　and　dealing　with　all　four　relations　at　the　same　time.　Moreover,　dynamic　threshold　is　brought　it?　BDOS　to　control　path　expansion　iteration.　After　experiments,　we　make　a　statistic　analysis　and　comparison　between　these　two　algorithms　and　another　approach　which　is　proposed　earlier　for　measuring　semantic　similarity　by　researcher　named　Yang.　The　experiments　show,　that　BDOS　gives　a　better　performance　and　accuracy.