Welcome to the CoBase research group. The following is a brief overview of the various projects.

      Cooperative database systems ( CoBase), under the support of the ARPA, and the direction of Professor Wesley Chu at UCLA's Computer Science Department, was started in the early 90s to provide knowledge-based approximate query answering to structure data (e.g. relational database). It has been successfully applied to transportation logistic planning systems for approximate matching the properties of objects (GLAD), medical image systems for approximate matching of medical image features (KMeD) and noisy emitter identification with Lockheed/Martin Marietta.

      Knowledge-based relaxation techniques have been extended to the XML data model ( CoXML) for information exchange and retrieval. CoXML cooperates with users and provides user specified structure and content relaxation. Evaluation with INEX test set yields effective and efficient retrieval performance.

      The knowledge-based approach has also been extended to free text in selecting scenario specific relevant query expansion terms to improve retrieval performance. Further, the knowledge-based approach has also been applied to semantic document ranking model (phrase VSM). These techniques have been used to improve retrieval performance for free text in the medical digital library project (KMeX).

      With the advancement of large scale web based complex information systems, security and privacy protection becomes increasingly important issues. The Inference Approach to Security and Privacy project (ISP) describes our recent progress in this area.

      Please feel free to browse around and we are happy to receive your comments.