Thesis Abstract and Dissertation Abstracts
Supporting Geo-recommendations In Location-based Community Services Based On Background Tracking Information
Abstract Category: I.T.
Course / Degree: M.Sc. in Computer Science
Institution / University: Technische Universität Berlin, Germany
Published in: 2012
Context-aware recommender system is a practical research theme in the field of ubiquitous computing. Through studying related works, it is found that context is often treated as a type of static and isolated information, which ignores the spatial and chronological continuity of context. Besides, the information source, upon which conventional recommender systems depend, might not be as reliable and constant as it was expected. Moreover, the sparsity problem and cold-start problem arise generally in recommender systems. In order to solve the problems, this thesis proposes a novel context-aware recommender system name Tracommender using background tracking information for providing location-based and community-based recommendation services.
Tracommender's information source is path and contextual information. Tracommender predicts a user's further movement by comparing the user's current path with a collection of historical paths called candidate paths. Collaborative filtering technique is employed to prepare the candidate paths according to user-location-visiting frequency. Two basic path matching approaches named adjacency matrix matching and minimum distance matching and two context-aware path matching approaches named context-weighted adjacency matrix and dimension-increased minimum distance are proposed to select path patterns from the candidate paths. As a solution to the cold-start problem additionally, a complementary location-based service named activity zone is proposed.
A prototype of Tracommender is implemented based on centralized system architecture comprising of a server, a database, and a client application on mobile devices. Finally, recommendation approaches and algorithms are evaluated. The conclusion indicates that the two basic path matching approaches complement each other well. The application of context-awareness in path matching approaches can help with improving recommendations, although a few influential factors have to be adjusted according to concrete cases.
Thesis Keywords/Search Tags:
location-based service, context-aware service, recommender system, tracking, path matching, activity zone, mobile advertising, recommendation
This Thesis Abstract may be cited as follows:
Wang, Y. (2012). Supporting Geo-Recommendations in Location-based Community Services based on Background Tracking Information. Master Thesis. Service-centric Networking, Technische Universität Berlin, Germany.
Submission Details: Thesis Abstract submitted by Yang Wang from Germany on 30-Jun-2012 14:30.
Abstract has been viewed 3532 times (since 7 Mar 2010).
Disclaimer
Great care has been taken to ensure that this information is correct, however ThesisAbstracts.com cannot accept responsibility for the contents of this Thesis abstract titled "Supporting Geo-recommendations In Location-based Community Services Based On Background Tracking Information". This abstract has been submitted by Yang Wang on 30-Jun-2012 14:30. You may report a problem using the contact form.
© Copyright 2003 - 2024 of ThesisAbstracts.com and respective owners.