Reconciling heterogeneous information sources

This paper discusses the results of experiments in automatic ontological reconciliation using AReXS, an implementation of the Example-Based Frame Mapping algorithm. Modications and extensions to the base algorithm are discussed in the context of comparing heterogeneous data from multiple on-line information sources. Consideration is given to the types of data encountered and their impact on the task of reconciling meaning. Other less automatic approaches to meaning negotiation are discussed and an argument is made for combining automatic and manual reconciliation techniques.