关系数据库关键词的纠错性查询及优化研究

发布时间：2018-03-01 07:03

本文关键词： 关系数据库关键词查询模式图候选连接树连接树　出处：《广西大学》2013年硕士论文　论文类型：学位论文

【摘要】：传统的关系数据库查询须指出将要查询的属性名,不能像搜索引擎那样自由地进行关键词查询。关系数据库关键词查询(KSORD)可以在不了解关系数据库结构的前提下,实现对关系数据库的自由态查询。本文对KSORD进行了分析和研究,针对以往KSORD的不足,提出了一个解决方案。本文针对以往KSORD缺乏关键词近似查询的问题,提出一个关系数据库关键词纠错性查询的解决方案。本方案给出一个高频词生成算法,利用该算法找出关系表中文本属性和字符串属性值的高频度词汇,并存储在hotkeys表中；给出一个关键词纠错算法,该算法依据hotkeys表中的热词,对待纠错的关键词进行纠错,得到纠错后的关键词。纠错后的关键词组通过KSORD进行查询,最终得到理想的查询结果。另外,本文针对以往KSORD查询效率低耗时长的问题,提出一个基于连接树表的解决方案,以缩短查询时间。在关系数据库模式图中,给图中各节点增加PageRank值,给出一个计算公式,为图中各边增加权重值；给出一个节点数不大于Maxsize的子图生成算法和一个子图中最小生成树对应的select模板生成算法。利用这两个算法从关系数据库模式图中生成连接树对应的select模板,并依据这些select模板建立连接树表joinTree。在查询过程中,从joinTree表中找到候选连接树对应的select模板,替代以往KSORD利用广度优先遍历算法(BFS)从模式图中找到候选连接树的低效方法。本文设计了关系数据库关键词纠错性查询的实验,以查询时间、查准率及查全率作为性能指标,验证了本文提出的关系数据库关键词纠错性查询解决方案的可行性与有效性；本文设计了基于连接树表的KSORD的实验,与以往的基于BFS策略的KSORD进行比较,在不降低查准率及查全率的情况下,缩短了查询时间,并且优化效果随关系数据库规模增大而更加明显。
[Abstract]:Traditional relational database query must point out the attribute name to be queried, and can not carry out keyword query as freely as search engine. The key word query of relational database / KSORD) can be done without knowing the structure of relational database. In this paper, KSORD is analyzed and studied, and a solution to the shortcomings of KSORD in the past is put forward. In order to solve the problem of lack of keyword approximate query in KSORD in the past, this paper proposes a solution for keyword correction query in relational database. The algorithm is used to find out the high frequency words of the text attribute and the string attribute value of the relational table and to store them in the hotkeys table. A key word correction algorithm is presented, which corrects the error correction keyword according to the hot words in the hotkeys table. The key phrase after error correction is queried by KSORD, and the ideal query result is obtained. In addition, aiming at the problem of low efficiency and long time consuming of KSORD query in the past, this paper proposes a solution based on join tree table to shorten the query time. In the schema diagram of relational database, we add PageRank value to each node in the graph. A calculation formula is given to increase the weight value for each side of the graph. A subgraph generation algorithm with no more nodes than Maxsize and a select template generation algorithm corresponding to the minimum spanning tree in a subgraph are presented. Using these two algorithms, the select template corresponding to the connection tree is generated from the relational database schema graph. The join tree table is established according to these select templates. In the process of query, the select template corresponding to the candidate join tree is found from the joinTree table, instead of the previous KSORD using the breadth-first traversal algorithm (BFS) to find the candidate join tree from the schema diagram. This paper designs the experiment of key word error correction query in relational database, taking query time, recall rate and recall rate as performance index, and validates the feasibility and effectiveness of the proposed query solution of key word correction in relational database. The experiment of KSORD based on join tree table is designed in this paper. Compared with the previous KSORD based on BFS strategy, the query time is shortened without reducing the recall rate and recall rate. And the optimization effect is more obvious with the increase of the scale of relational database.
【学位授予单位】：广西大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP311.13

【相似文献】