期刊
CONSTRUCTION MANAGEMENT AND ECONOMICS
卷 27, 期 3, 页码 241-251出版社
ROUTLEDGE JOURNALS, TAYLOR & FRANCIS LTD
DOI: 10.1080/01446190802714781
关键词
Ant colony algorithm; artificial intelligence; case-based reasoning; decision trees; litigation; neural networks
类别
The frequency of construction litigation has increased over the years, making litigation a costly and timeconsuming activity. It is in the interests of all parties to a construction contract to avoid litigation. A tool (Ant Miner) is proposed to predict the outcome of construction litigation, hence encouraging the parties to settle out of court. Ant Miner, a rule-based classification system extracts classification rules by using ant colony optimization. It is used on 151 Illinois circuit court cases filed in the period 1987-2005. The prediction model is composed of data consolidation, attribute selection, classification and assessment. The results provide evidence that Ant Miner performs better than models used in earlier studies and that the rule sets discovered by this tool are highly interpretable, but that this tool suffers a great deal from noisy data. If the parties involved in a dispute have access to the proposed system that predicts the decision of the courts with higher accuracy and reliability than before, then they are expected to avoid litigation and settle out of court in order to save considerable time, money and aggravation.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据