Account correlation method used for UGC (User Generated Content)-spanning website platform

The invention discloses an account correlation method used for a UGC (User Generated Content)-spanning website platform and belongs to a method for correlating accounts of one entity user. The basic principle is that the plurality of accounts belonging to one entity user on different UGC website platforms are correlated by extracting characteristics from text content generated from UGC website accounts. The method comprises the following steps of obtaining data, pre-processing the data, extracting the characteristics and filtering layer by layer. The data obtaining step is used for collecting the text content generated from the user accounts of a target UGC website. The data pre-processing part is used for pre-processing the text content. The characteristic extracting step is used for extracting sex characteristics, age characteristics, geographic position activity characteristics and writing style characteristics from the text content respectively. The step of filtering layer by layer is used for filtering accounts which are not related to the given user accounts through the sex characteristics, the age characteristics, the geographic position activity characteristics and the writing style characteristics layer by layer in sequence. By the aid of the account correlation method, the problem that the accounts of one entity user on the different UGC websites are not related can be effectively solved and the practical value is very high.