Recommender system for targeted advertising

Advertising in mobile apps are increasing and so is the need to show the right ad to the right user. This study was conducted in cooperation with Seekly AB, a company whose app displays a feed with upcoming events in a users immediate area. In this app, every event is associated with an interest and users can choose interests to follow in a list. Seekly wanted to use so called behavioral targeting to show ads in their feed. The solution that was developed is useful for all Seekly-like systems and consists of a content based recommender system that chooses ads based on the interests a user has selected and events and ads that the user has shown an interest in. Apps that in some way or another resembles the Seekly app are not uncommon and recommender systems for behavioral targeting suited for this kind of system are to the best of our knowledge not described in the literature. The resulting recommender system has been implemented and shown to be able to recommend ads that has been associated with interests that match the interests selected by the user and/or amplified by his or her behavior. There are also indications that the system would be able to increase the number of ad clicks compared to randomly selected ads, but no statistically significant proof was found