A b stra ct A b s t r a c t W ith the rap id d eve lop m ent o f and e-co m m erce,in fo rm atio n o v erlo ad in g beco m e s a serio us p ro b lem fo r th e e-co m m erce u ser . M ore an d m ore produ ct in form atio n gen erated in e-co m m erce m ak es it d ifficu lt for u sers to qu ick ly fin d their fa v orite p rod t the sam e tim e, e-co m m erce co m p an ie s are faced w ith th e p ro b lem th at ho w to p ro v id e u sers w ith pro d ucts in tim e to im p ro ve sales. R eco m m en d atio n sy stem is an effective w a y to so lve such pro b lem .C o llab orative F iltering is on e o f the best an d w id e ly u sed tech no lo g ie s in reco m m end ation system ,it is also fac in g d ata sp arsity, co ld start, sca lab ility a n d o th e r is su e s . T h is article p ro po ses a T rust—b ased C o llabo rativ e F iltering a lg o rith m to so lve the data sp arsity p ro b lem o f the trad ition a l C o llab o rative F ilterin g alg orithm and im p ro ve the reco m m e ndatio n accuracy. T ru st can d escribe the re latio nsh ip b etw ee n users m o re accu rate ly ; tru st p ro p agatio n can co n n ect u sers th at o rig in a lly w ith o ut interactio n , use tru st a s th e sup p lem ent o f sim ilarity ca n alleviate the d ata sp arse p rob lem o f trad itio n a l co llab orative filtering re co m m en datio n a lg o rith m and im p ro v e th e p red ictio n accu racy o f re co m m endatio n sy stem . T h is artic le d esig ns 3 exp erim ents u sin g the c lassic data sets to verify th e v a lid ity o f th e a lgo rith m . E x p erim ents sho w that T ru st-b ased C o llaborative F iltering a lg o rith m can a llev iate the d ata sp arsity pro b le m an d im pro ve the reco m m end atio n accu racy. A t last, the article u ses th is algo rith m in A b s tra ct the d e sig n o f、reco m m end atio n m o d u le o 「 am o n lin e bo o k sto re fo r bo o k re c o m m e n d a t io n . K ey w o rd s E -C o m m erce , R eco m m e n d atio n S y st