COST

The purpose of this work stream is to explore how structured expert judgment alongside data mining can help to structure and elicit parts, or all, of a directed acyclic graph (DAG) and to explore the sensitivity of the output of analysis to this structure.  We hope that the following ideas will be explored during the COST network:

a.       Develop and validate data mining tools that help to learn and structure DAGs in combination with structured expert judgement.

            i. Use structured expert judgement to help to explore the identification of false positives identified when mining large datasets

             ii. Use data mining to support the development of dependencies when challenges exist with eliciting expert judgement

b.      Identify and develop a standard procedure to structure DAGs based on current academic literature and an evaluation of current and best practice

c.       Develop tools to support remote and “online” elicitation of DAG structure.

The expected outputs of this work stream is clear tools and processes to support practitioners and academics that utilize DAGs on a regular basis. The development and dissemination of these tools and processes will generate evaluation and feedback from the user community and help to refine current thoughts.