Retrieving similar cases for alternative dispute resolution in construction accidents using text mining techniques

Abstract

The occurrences of construction accidents can be reduced but cannot be avoided because of risky and uncertain nature of construction works. Construction accidents tend to cause serious damages, injuries, fatalities, and work interruptions, and provoke disputes among the involved parties. One useful means for resolving the dispute through alternative dispute resolution, instead of costly litigation, is to find similar accidents occurring in the past in the same jurisdiction and compare their similarities and differences. This paper introduces a research project on the effective retrieval of relevant historical cases from a case library using text mining techniques. Research issues in the text mining process, especially how to represent the unstructured textual cases by a structured term vector model, are investigated. It is concluded that natural language based case document retrieval is superior to the case-based reasoning from structured case collection and is more practical for implementation in a construction management information system.