| dc.description.abstract | As the available content on the Internet grows, it is becoming harder to find the relevant data you are looking for. A segment of the available content is future events (e.g. concerts, sports matches, public parties, book releases, cinema screening events, etc.) If a user wants to know what is happening next weekend in the nearest city, or is planning a vacation, he/she usually want to get a ranked list of personalized recommendations rather than scrolling through dozens of pages finding something interesting.
There is little relevant academic research done in this field in particular. The Thesis begins by briefly describing the state-of-the art, then presenting some already existing future event recommender systems and sources that provide Norwegian future events.
The system described in the thesis aims to extract Norwegian screening events and its relevant features. The data is modeled using JSON-LD with the Schema.org vocabulary. Then use this data to perform personalized recommendations. Two surveys were conducted to gather user preferences and find importance of relevant features. A simple recommender system were created, testing the performance of different features for a content-based filtering recommender. A collaborative filtering and a weighted hybrid recommender were created as well. And finally comparing the three approaches. | |
