摘要
给出了最小可查询模式MEP的概念,并在此基础上提出了MEP生成算法与基于MEP的自适应查询方法。该方法将查询接口由单文本框推广到最小可查询模式集,一次查询由一个MEP和与该MEP匹配的关键词向量共同确定,自适应地产生期望最优的下一个查询,直到满足查询停止条件。该方法克服了当前Deep Web查询方法能力不足导致的"数据孤岛"问题。在6个实际Deep Web站点的实验表明,该方法比已有方法具有更强的查询能力与适用性。
This paper proposes the concept of minimum executable pattem（MEP）, and then presents a MEP generation method and a MEP-based Deep Web adaptive query method. The query method extends query interface from single textbox to MEP set; it performs a query by choosing a MEP and a keyword vector of the MEP, and generates the next expected optimal query until stop condition is satisfied. The proposed method overcomes the problem of ＂Data Island＂ which results from deficiency of current methods. The experimental results on six real-world Deep Web sites show that our method outperforms existing methods in terms of query capability and applicability.
基金
国家自然科学基金(60825202
60803079)
国家高技术研究发展计划(863计划)(2008AA01Z131)
新世纪优秀人才支持计划(NECT-08-0433)
高等学校博士学科点专项科研基金(2009021110060)
