Enabling custom application content through semantic web filters

摘要

Understanding user intent is critical as we move towards provisioning next generation services. Towards this goal, we have developed a mobile in-device framework that mines user consumed content for inferring custom user interest. We build custom filters from user content that has wide ranging applications including filtering content automatically from applications employing web based data backend. Our framework is based on latent extraction of content based on a hybrid supervised and unsupervised topic modeling process. The extracted filters are coupled to popular web based services for search and retrieval of content backed by a unique UX applicable for mobile systems. Our trial results show a subjective satisfaction score of over 80% for filter results along with 94% precision score for supervised latent topic-based content inference.