Due to the ever increasing need for information, it is vital to stay up-to-date with emerging events. Common sources of information are news messages, which contain events that could be of importance for people. Hermes is an ontology-based framework for building news personalization services, which focuses on news classification and knowledge base updating. Furthermore, the framework allows for news querying and result presentation. In this paper, we focus on the techniques involved in keeping Hermes' internal knowledge base up-to-date. Essentially, our semi-automatic approach to knowledge acquisition from news is based on ontologies and lexico-semantic patterns. Experiments with an implementation of the framework, the Hermes News Portal (HNP), show for news classification a precision and recall of 86% and 81%, respectively. Usability tests demonstrate that user interaction with the system, needed for knowledge base updating, is positively assessed by the users.
|Number of pages||2|
|Publication status||Published - 25 Oct 2010|
|Event||Twenty-Second Benelux Conference on Artificial Intelligence (BNAIC 2010) - Luxembourg, Luxembourg|
Duration: 25 Oct 2010 → 26 Oct 2010
|Conference||Twenty-Second Benelux Conference on Artificial Intelligence (BNAIC 2010)|
|Period||25/10/10 → 26/10/10|