Event Recommendation Based On User’s Interest And Geographic Location

With the social network sweeping the world, all kinds of SNS sites have also sprung up quickly.People are no longer satisfied with online activities and interaction, but turning to offline activities.Meetup website has been known for community service.Through some groups, some people with common interests are organized to participate in the offline activities. In order to let users better enjoy the fun, event-recommendation algorithms have become a hotspot. Traditional recommendation algorithms are mostly based on users’ interests and labels, but ignoring diversified social relations and location information. In this paper, the model combines the traditional recommendation algorithm, taking into account the user’s personal preference, a variety of social relations, the location of the heterogeneity and activity content. An event recommendation based on user’s interest and geographic location is proposed. It combines the latent factor model with logistic regression model to improve the accuracy and reliability of the recommendation algorithm.The model firstly recommends the groups, and then recommends the events on the base of group recommendations. It will narrow the recommended range and improve the efficiency. As for the assessment of the model, we use the value of the area under the curve AUC and value F to evaluate the model.In order to illustrate the superiority of the model, we compared the event recommendation which based on user’s interests and geographic location with the social regularization recommendation and another event recommendation based on collaborative filtering.Then its recommended results will be applied to the user’s personalized recommendation.The experimental results show that the event recommendation based on user’s interest and geographic location is better than the other two models.