Blog friend recommendation method based on tree log pattern analysis

The invention discloses a blog friend recommendation method based on tree log pattern analysis. An offline digging method is adopted, a server log is analyzed to extract the access record of a visitor on a blog page; an access log tree which takes a blog to be recommended as a root is further constructed by the technologies of grouping, sorting, loop-back removal and the like; the constructed access log tree is frequently dug to find a frequent subtree conforming to a preset requirement; nodes in the frequent subtree serves as candidate blog friends; a recommendation degree is calculated according to the set formula; and friends with highest scores are recommended. An algorithm in the method is different from the traditional algorithm based on frequent item digging or frequent sequence digging; a frequent tree structure digging method is adopted for fully digging aiming at the specific parallel link relationship and indirect access characteristic of the blogosphere; and potential access contact among blogs is extracted and is recommended to an access user, thereby improving user experience. The blog friend recommendation method is an efficient and practical blog recommendation method.