AKT Player - SELLIC

AKT Player URL:

http://www.sellic.ed.ac.uk

AKT Player POC:

John MacColl <john.maccoll@ed.ac.uk>

AKT Team POC:

Austin Tate <a.tate@ed.ac.uk>

Area of Interest:

• How can scientific researchers use the semantic web in order to discover newly published knowledge precisely relevant to their domains?
• Statement of interest from John MacColl on 18-Sep-2001:
  

  My particular focus in SELLIC at Edinburgh is agent-based research information. I am hoping to put together a project - perhaps for EC IST (semantic web technologies) funding (or other?) to create an agent or agents which harvest specifically tailored research information for the needs of scientific researchers from a wide range of commercial and non-commercial web content sources. These sources will consist of data both in text form (eg research papers) and structured data (eg gene sequence datasets), and the agent should be capable of relating text to strucutred data. I am interested in using OIL in this project, and have discussed with Dieter Fensel at VU Amsterdam the roles which OIL could play in source annotation, interchange formats and agent competence description. We reckon we need one partner to do tool development, and case study partners likely to be content providers - both commercial and non-commercial (I have two already in mind: a major science/technology/medicine academic journal publisher/aggregator and another database/journal aggregator).

  Any advice or expressions of interest from the AKT group would be welcome. My own background is in librarianship/information science, so the role of SELLIC would be in coordination, user analysis, testing and evaluation.

Relevant Work Areas:

• To develop agents which can retrieve new knowledge from web/grid which conforms to a dynamically updated researcher profile. Sources encompassed by such agents to include all relevant container types (eg web sites of fellow researchers; preprint archives; commercial sites such as journal publishers and aggregators, etc) and all object formats (including pdf, postscript etc)
• To develop systems for the automatic generation of metadata from relevant ontologies and resource description formats (eg bibliographic metadata)
• To harmonise the knowledge technology effort with that of digital library developments, particularly in the following areas:
  • Presentation: how is the system meshed with library systems? The library's range of sources - including those commercial data services for which it has licensed access - need to be included in the target set. The results need to be presentable via a common interface: information visualisation holds promise here.
  • Interface with human editors: human editors (eg library cataloguers) will be needed for some time for metadata quality control in the quest for precision. How can systems make decisions on what to refer to human editors (eg in the case of multiple subject descriptors), and how can they manage this workflow? They need to analyse usage of retreived information in order to be able to refine search strategies. This analysis should be more sophisticated than simple click-through assessment - eg can the user be asked questions by the system which help it refine the strategies?

Interest in AKT Challenges:

1. Acquistion:
2. Modelling:
3. Re-use:
4. Retrieval and Extraction:
5. Publishing:
6. Maintenance:

Interest in AKT Testbeds and Scenarios:

Other Joint Work:

1. ESPRIT proposal for research on ontologies to underpin scientific publishing in the environmental sciences (SELLIC - John MacColl; University of Edinburgh Environmental Sciences - Bob Muetzelfeldt; AKT - Stuart Aitken)