Warning: your browser seems to be quite old, or at least old enough that it does not support the minimal standard capabilities of today's browsers, which are used extensively on this site. Please upgrade to a browser that supports web standards, specifically one that supports minimal CSS1 capabilities, such as Netscape 6 or higher, IE 5 or higher, or Opera 5 or higher. Your browsing experience will be much better! Thanks.
Title Automatic paradigms for semantic annotation.
Starting date October 2005.
Ending date April 2008.
Coordinator Prof. B. Retjin (University of Belgrade)
WG5 carries out a collaborative research on semi-automatic paradigms for semantic annotation, focusing on:
- The use of relevance feedback for semi-automatic creation of high-level metadata and correction of conceptually wrong descriptors generated automatically. This work is based on the study of immune system paradigms and schemas based on evolutionary algorithms to create predictors and the use of these predictors to guide future queries and learning principles.
- Automatic generation of semantic metadata from low-level descriptors and available semantic annotated content using inference models and evolutionary learning. It is assumed that retrieval gets started with a set of seed images and sounds or music files containing descriptors generated automatically by other algorithms. The user can accept or reject the retrieved document. Using this positive and negative user response the system is fed and a new search is started. The WG is exploring the use of genetic algorithms, of immune system paradigms, and of other techniques based on learning a predictor and then using the predictor to guide the future queries. A central objective is to minimise the number of documents the user has to grade (retrieval iterations).