Information Retrieval & Reuse

This research has developed innovative digital technologies for building and applying intelligent information systems for real-world problems, in particular decision support, product design, project planning, recommender systems and information management.

The main focus is on the retrieval and reuse of information using case based reasoning, where problems are solved by retrieving and reusing similar experiences that are stored in a case base.

Research has developed knowledge-light learning techniques that use only the existing cases to learn explicit knowledge to improve the design and maintenance of the case-based system. Novel complexity-based editing tools have been effective in discovering new cases, reducing redundancy and removing faulty cases. A software tool has been developed that transforms a database of cases into a fully-fledged case-based system including optimised retrieval and reuse.

IR2 Image 1

A suite of text mining software components has been developed to extend the open-source jCOLIBRI framework. Concept discovery algorithms establish similar context within textual content. Semantic knowledge is extracted from concepts and organised into domain-specific taxonomies. Recent projects have turned their attention to multimedia data for image management and music recommendation. Information interaction is also used to incorporate usage into the indexing of multimedia cases.

Information retrieval research has focused on information seeking. A document collection model has been developed that summarises the underlying knowledge structures and represents a user’s search process and space. Research on context-sensitive Information retrieval has explored “entanglements” between users, textual and multimedia content and retrieval contexts.

Industrial partners include: Selex Galileo, ESA, AstraZeneca, AxSys.

Academic collaborations involve: IIT Madras, Universidad Complutense de Madrid, Northern Research Partnership (NRP), Scottish Informatics & Computer Science Alliance (SICSA), Scottish Academy of Fashion (SAF).

Current Projects:

AHRC - STAR Accessing Implicit Knowledge of Textiles and Design (AHRC with Heriot-Watt & Johnstons)
Living History (Horizon with Historic Scotland & AmbieSense)
Case-Based Workflow Creation and Adaptation
Knowledge Extraction from Text (SICSA)
Content-Based Video Retrieval
Hybrid User Profiling and Adaptation (NRP with Yahoo!)
Audio Feature Extraction for Music Recommendation
Context-sensitive High-order Language Model (RSE/NSFC with Tianjin)
QONTEXT Information Access and Retrieval (EU FP7 Marie-Curie)
AutoAdapt for Assisted Information Seeking (EPSRC with Essex & OU)
Context-Sensitive Information Retrieval (EPSRC with Glasgow & OU)