The invention provides an apparatus, a method and a computer system for content recommendation and a recording medium. A base/inversion component extractor calculates an occupancy rate of each component of a vector of user preference information obtained from information stored on a user preference database, and extracts a base component from the user preference information in accordance with thecalculated occupancy rate of the base component. The base/inversion component extractor extracts a similar base component from item preference information obtained from information stored on an item metadatabase. A recommendation engine calculates a similarity between a base vector of the user preference information and a base vector of the item preference information, and identifies, as candidate items, items of the predetermined number in the order of from high to low similarity. The recommendation engine further calculates a similarity between an inversion vector of the user preference information and an inversion vector of the item preference information, and identifies an item candidate having a low similarity.