Content-Based Retrieval (CBR) System for Ottoman Archives

We propose a content-based retrieval (CBR) system for digital Ottoman archive documents. In this system, the symbols extracted from the documents are matched with the most similar one in the symbol library, which is created in a supervised manner. The users specify queries by marking a region on an example document and the system retrieves all documents that include the symbols found in the query region. A prototype of the system is currently available on the Web