- Keynote Speakers
Professor Stephen RobertsonMicrosoft Cambridge, UK
Title:The study of information retrieval - a long view
This talk will include a rapid canter through the roughly 50-year history of IR system evaluation, beginning with Cranfield. A major theme will be the opposition between, on the one hand, experiments conducted in a laboratory setting, and on the other, studies of real people with real information needs. This will lead on to a discussion of the compromises that are made in pursuit of scientific validity and reliability. It will be argued that such compromises are necessary, but also that variety is necessary too. We need a range of experiments, investigations and studies on the spectrum from real-life to full laboratory control; we also need to approach the isolation and abstraction required for laboratory experiments in different and complementary ways. Finally, I will argue for a scientific approach to theory, and against effectiveness measurement. downloadBio:
Stephen Robertson has been researching in the field of information retrieval for a while, beginning when he took his MSc at City University in 1968. He has approximately 90 published papers, mostly in the modelling and evaluation of IR systems. The ranking algorithm known as (Okapi) BM25, which is currently regarded as state-of-the-art, emerged from 20 years of work on the probabilistic model for IR, with (variously) Karen Sparck Jones, Keith van Rijsbergen and Stephen Walker. He led the group at City University which took over development of the Okapi system and used it for a series of user experiments. His career includes a spell at the Aslib Research Department, PhD and a research fellowship at University College London, 20 years at City University (with a short spell at the University of California Berkeley), and now ten at Microsoft Research in Cambridge. He retains his City University professorship, as well as being a Fellow of Girton College Cambridge. He received the Tony Kent Strix Award in 1998 and the Gerard Salton Award in 2000.
Dr. Ian RuthvenUniversity of Strathclyde, UK
Title:The context of the interface
Our ideas on context (from both a soft and hard laboratory perspective) often manifest themselves at the interface. Interfaces for information seeking range from systems which offer little contextual information, or flexibility of use, to systems which are highly driven by the searcher's individual interaction. In this talk I will use existing search interfaces to highlight how the nature of information retrieval interfaces has changed in response to research from contextual IS&R. A particular theme will be the commercial nature of web search interfaces and how approaches such as persuasive and emotional design can be used to encourage interaction. I will also consider the nature of specialised versus general purpose search interfaces and the challenges raised in interface design. The (not very) hidden agenda behind this presentation will be to argue for an interaction-centred approach to IR systems. downloadBio:
Ian Ruthven is a Reader in Information Seeking and Retrieval at the University of Strathclyde. At Strathclyde he heads the i-lab research group, a broad inter-disciplinary research group whose research spans statistical data modeling, information retrieval, digital libraries, mobile information access, information strategy and public libraries. His research interests are in the broad area of interactive information access. This includes modeling of interactive retrieval systems, user and technical evaluations, interface development and studies of systems in use. His most recent research has examined areas such as Personal Information Management, Complex Interactive Question Answering and the evaluation of novel surrogates for interactive information retrieval.