Argument Representation and Conceptual Retrieval for Litigation Support


Author: Paul Hosking
Source: GZipped PostScript (347kb); Adobe PDF (1176kb)

Case preparation and evidence management are fundamental tasks faced by litigators. These tasks are not well supported by existing litigation support software. Case preparation involves constructing and evaluating arguments in support of the client; current software packages provide only limited assistance, as they are unable to represent arguments. Evidence management involves organising and retrieving documents associated with the case; existing document retrieval systems employ only crude keyword-based indexing schemes that achieve poor rates of recall.

This dissertation describes a prototype system that uses an explicit representation of arguments to enable lawyers to construct, edit, browse, and evaluate arguments graphically. The representation includes a well-defined syntax and a semantics characterising the degree of support for an argument. Argument construction is facilitated by the provision of a set of argument packets encoding standard patterns of inference, which may be retrieved and copied into arguments as they are constructed. Browsing is facilitated by a set of argument views that enable the user to focus on specific portions of the argument. Argument evaluation is supported by a further set of argument views that evaluate the level of support for a given fact in the argument and highlight areas of weakness. Documents associated with cases are stored and retrieved in an innovative manner, using the structure of the argument as an index to a hypertext database.

[Up to Computer Science Technical Report Archive: Home Page]