Application recommendation method and server

Abstract

A method and apparatus are provided. The method includes prior to receiving a current search term from client application of a user among plural users, obtaining historical search terms entered by the user during a first time period and, for each historical search term, one or more historical applications accessed by the user after entering the historical search term, separately obtaining feature data of the historical search terms, the one or more historical applications and candidate applications, the feature data being data combined from all of the plurality of users, training the feature data according to a mining model, and obtaining through mining using the trained feature data, for each candidate application, a recommendation term corresponding to the candidate application. In response to receiving the current search term from the client application, one or more applications from the candidate applications are determined according to the current search term and the recommendation terms, and transmitted to the client application.