Web数据库top-k多样性关键字查询推荐方法

发布时间：2018-01-21 09:14

本文关键词： Web数据库多样性推荐耦合关系典型化分析 top-k选取　出处：《计算机研究与发展》2017年07期 　论文类型：期刊论文

【摘要】：Web数据库用户通常使用他们熟知的关键字表达查询意图,这可能导致获取的结果不能很好满足其查询需求,因此为他们提供top-k个与初始查询语义相关且多样化的候选查询有助于用户扩展知识范围,从而更准确完善地表达其查询意图.提出一种top-k多样性关键字查询推荐方法.1)利用不同关键字在查询历史中的同现频率和关联关系评估关键字之间的内耦合和间耦合关系;2)根据关键字之间的耦合关系构建语义矩阵,进而利用语义矩阵和核函数方法评估不同关键字查询之间的语义相关度.为了快速返回top-k个与初始查询相关且多样性的候选查询,根据查询之间的语义相关度,利用概率密度函数分析查询的典型程度,并利用近似算法从查询历史中找出典型查询.对于所有的典型查询,从中选出少数代表性查询,根据其他典型查询与代表性查询之间的语义相关度,为每个代表性查询构建相应的查询序列;当一个新的查询到来时,评估其与代表性查询之间的语义相关度,然后利用阈值算法(threshold algorithm,TA)在预先创建的查询序列上快速选出top-k个与给定查询语义相关的多样性候选查询.实验结果和分析表明:提出的关键字之间耦合关系计算和查询之间的语义相关度评估方法具有较高准确性,top-k多样性选取方法具有较好效果和较高执行效率.
[Abstract]:Web database users usually use keywords they know well to express their query intent, which may result in the obtained results not being able to meet their query requirements. Therefore, it is helpful for users to provide top-k with a variety of candidate queries related to the semantics of the initial query. In order to express its query intention more accurately and perfectly, this paper proposes a top-k diversity keyword query recommendation method. 1). The cooccurrence frequency and correlation relation of different keywords in the query history are used to evaluate the coupling and coupling between keywords. 2) constructing semantic matrix according to the coupling relation between keywords. Then the semantic matrix and kernel function are used to evaluate the semantic correlation between different keyword queries. In order to quickly return top-k candidate queries related to the initial query and diversity. According to the semantic correlation of the query, the probability density function is used to analyze the typical degree of query, and the approximate algorithm is used to find out the typical query from the query history. A few representative queries are selected and corresponding query sequences are constructed for each representative query according to the semantic correlation between other typical queries and representative queries. When a new query arrives, the semantic correlation between the query and the representative query is evaluated, and then threshold algorithm is used. ). Top-k multiple candidate queries related to the given query semantics are quickly selected from the pre-created query sequences. The experimental results and analysis show that:. The proposed method for evaluating the semantic correlation between the key words and the query has a high accuracy. Top-k diversity selection method has better effect and higher execution efficiency.
【作者单位】：辽宁工程技术大学电子与信息工程学院;辽宁工程技术大学软件学院;
【基金】：国家自然科学基金青年科学基金项目(61401185) 辽宁省自然科学基金项目(20170540418) 辽宁省教育厅科学技术研究项目(LJYL018)~~
【分类号】：TP311.13;TP391.3
【正文快照】： 关键字查询无需用户了解Web数据库的结构和内容,而是类似于谷歌和百度等搜索引擎一样仅使用少数几个关键字表达查询意图.近年来,关系数据库上关键字查询的代表性研究工作主要是基于模式图(schema graph,SG)[1-3]和候选网(candidate networks,CNs)[4-6]的全文匹配方法.然而,上

【参考文献】