论文部分内容阅读
针对已有查询接口匹配方法匹配器权重设置困难、匹配决策缺乏有效处理的局限性,提出一种基于证据理论和任务分配的DeepWeb查询接口匹配方法.该方法通过引入改进的D-S证据理论自动融合多个匹配器结果,避免手工设定匹配器权重,有效减少人工干预.通过对任务分配问题进行扩展,将查询接口的一对一匹配决策问题转化为扩展的任务分配问题,为源查询接口中的每一个属性选择合适的匹配,并在此基础上,采用树结构启发式规则进行一对多匹配决策.实验结果表明ETTA-IM方法具有较高的查准率和查全率.
In order to overcome the limitations of traditional methods such as the difficulty of setting the matching weight of matching interface and the lack of efficient processing of matching decision, a deep Web query interface matching method based on evidence theory and task assignment is proposed. This method introduces an improved DS evidence theory A mathematicator can avoid manually setting the weight of matcher and reduce manual intervention effectively.Based on the extension of task assignment problem, the one-to-one matching decision problem of query interface is transformed into an extended task assignment problem, Based on this, the tree structure heuristic rules are used to make the one-to-many matching decision.The experimental results show that the ETTA-IM method has high precision and recall rate.