THE METHOD AND APPARATUS FOR IMAGE RECOMMENDATION BASED ON USER PROFILE USING FEATURE BASED COLLABORATIVE FILTERING TO RESOLVE NEW ITEM RECOMMENDATION

摘要

PURPOSE: An image recommendation method based on user profiles which combines contents based filtering and collaborative filtering and a device thereof are provided to reduce the decrease of recommendation result accuracy when input data about client preference is insufficient. CONSTITUTION: Using a specific vector of a multi dimensional property space, entire background images of a client purchase list database are partitioned into multiple meaningful areas(1~n). Properties are extracted from the areas of the partitioned background images. A user profile is generated. Using multimedia image contents of a target user and other arbitrary users, a similar group is set. A background image recommendation list is generated.