The topic of my Master thesis
Sunday, 04/26/2009, 11:19:38
Sorry for not posting lately, I was busy finding and preparing my master thesis topic. Here, I would like to introduce it to you:
Incorporating contextual information in recommender systems
Today, people are confronted with a large amount of information in the World Wide Web. Receiving a small subset of desired and filtered content through standard search engines turns out to be very difficult. And it becomes quite impossible, if user specific needs and interests should be taken into consideration.
Recommender systems handle this issue by providing personalized content recommendations to the user based on his personal background, preferences and interests.
Numerous recommendation methods were designed over the years to enhance the preciseness of recommendations, such as content-based and collaborative filtering or hybrid approaches.
Even though hybrid methods often deliver acceptable results, incorporating contextual information into the recommendation process may be very important to get more precise results for the situation the user is currently in.
Current recommender systems as described above primarily focus on recommending items to users and vice versa. Existing ratings for items are the basis for effective recommendations. In many modern, mobile applications however, it may not be sufficient to consider only users and items. Contextual information may provide a significant preciseness to the recommendation.
If for example, a user is vegan, eats only organic and healthy food, goes shopping nearby and tries to live economical, it would not make sense to recommend him stores far away or only discounters without taking his preference for vegan food into consideration.
The objective of this thesis is to create a recommendation algorithm using the "SMART Recommendations Engine" of Fraunhofer FOKUS that incorporates contextual information to deliver more precise results taking users’ current situation and preferences into account.
Furthermore a mobile application is to be implemented to demonstrate the recommendation algorithm in a predefined scenario.
Useful papers:
- Multidimensional Recommender Systems: A Data Warehousing Approach
- Incorporating Contextual Information in Recommender Systems Using a Multidimensional Approach
| Author: Abdulbaki | [no comments] |