ABSTRACT
This chapter proposes a Bayesian Network based mining framework utilising product model data as a primary source of engineering knowledge to support domain-specific information retrieval as well as re-organisation and re-contextualisation of document information. Numerous standardised document schemata have been developed for exchanging machine and at the same time human readable information e.g. for bidding, tendering and controlling. For a flexible and sustainable reuse of information supporting ad-hoc location of information in documents for operations knowledge management, it is necessary to utilise additional context information and appropriate background knowledge. Bayesian Network based Information Retrieval models associate index terms, documents, user queries and user’s information needs with random variables to model the retrieval task as an evidential reasoning process. The product model and the concept model layer together represent the configurable knowledge model used to trigger and control Information Retrieval and mining processes. The first extension to the mining network accounts for term interdependencies in natural language text.
