feature selection for anti-cancer plant recommendation

Every year tremendous experimental analysis has been done for evaluation ofanti-cancer properties of plants. a good ranked list of potential anti-cancer plants whichraised out of veried anti-cancer metabolites, reduces the time and cost for evaluatingplants; otherwise, we charged for testing unrelated plants. ranked list produced byanalyzing plant-metabolite biological graphs are candidate for such situation. graphnodes are ranked according to some graph features. a problem with this approach ishow to select the good features of graphs. in this paper a metric used in informationretrieval and recommender systems is employed for comparing two dierent ranked list.in an information retrieval system such as search engines, a good system should showthe top results rst. a metric named average precision is used here for discriminatingdierent lists, resulted from dierent features. we build a network of similarity ofplants according to their common metabolites. after that, with various combinationsof the graph features, the plants are ranked. the subset of features which produces theranked list with higher ap score is considered as the best features for anti-cancer plantrecommendation. the proposed method could be employed to select the best graphfeatures in screening of anti-cancer plants from an unveried plants list. so that, theplant with higher score in the list have higher chance to have anti-cancer properties.