- Kevin Ashley: Progress in Text-Based Case-Based Reasoning
In recent years, CBR researchers have become increasingly interested in dealing with textual representations of cases. Many CBR applications involve the need to handle cases available as texts. The case texts may be only partially structured or completely unstructured, in stark contrast to the highly structured cases and case representations of more traditional CBR systems. This talk identifies some of the major problems to be solved, explains how these problems distinguish Text-Based CBR from other areas, such as Information Retrieval, and surveys some of the pioneering efforts to solve them. In particular, The talk addressed such topics as: How can texts be employed in a case-based reasoning system? What kind of knowledge is required to build a text-based CBR system? How can application-specific knowledge be used to assist in assigning indices to cases as texts? What roles can traditional structured case representations play in helping to assign indices to textually-represented cases? How should Text-Based CBR be evaluated?
- Pádraig Cunningham: Intelligent Systems in Electronic Commerce
There are two key motivations for intelligent systems support in e-commerce. The first is the need to provide software sales assistants to lubricate the sales process in the absence of human sales support. This motivation envisages the replication of an existing sales process online. The second motivation is to make use of the data made available because of the electronic nature of the interaction. This second motivation envisages a new type of sales process not possible outside the online context. This talk looked at how CBR can be used to build online sales assistants and at the new possibilities for intelligent support in the online sales that exploit the data that is available.
Slides as PDF file (687kB)
- Boi Faltings: Case Adaptation - New Perspectives for Case-Based Reasoning
Traditionally, CBR has focussed on case retrieval and regarded adaptation as a secondary, sometimes annoying problem. However, in many domains - architecture, business, but also word processing - case adaptation, and not retrieval, is where problem-solving is required. The speaker argued that case adaptation by itself can be a useful new metaphor for interactive problem-solving tools. He showed examples of prototype design tools that illustrate such a possibility, and discussed some of the issues which are important for making it a mainstream technology.
Slides as PDF file (56kB)