Information recommendation system adapted to cloud advertising service

The paper considers the problem of information overload of the users of large-scale Internet services. The amount of data in the global information space is growing much faster than the ability of an individual user or client to handle them. As a result, information overload has become a serious problem for all Internet services. The solution of the problem is the implementation of the recommendation system. The methods of the recommendations in the related subject areas are described, as well as adaptation of these methods to the domain of advertising services is made. Two categories of the users are defined: advertisers — companies that wish to order advertising services and distributors — people who are ready to provide this advertising service. The main objective of this internet-service is to make as many transactions between advertisers and distributors as possible. The objects of the recommendations in this system are users. The analysis of selection of the users’ criteria is made; definitions of explicit and implicit criteria are provided. The evaluation function of correspondence of the users, which is based on the algorithm of vector spatial modeling, is considered. The algorithm enhancing the quality of the recommendations by taking into account the interests of the recommended and recommended sites, which in their turn partially solve the problem of information overload of the users of the resource is presented.