CAJ | 학술논문

提出了一种新的自适应结构索引:AS-Index(adaptive structural index),能够克服现有静态索引和自适应索引的缺陷,具备高效的查询和调整性能.AS-Index建立在F&B-Index的基础之上,其索引结构包括F&B-Index,Query- Table和Part-Table.Query-Table能够记录频繁查询,避免了查询过程中的冗余操作.并且,在Query-Table的基础上提出了自底向上的查询处理过程,能够充分利用现有的频繁查询高效地回答非频繁查询.Part-Table用于优化包含祖先后裔边的查询,进一步提高了查询性能.现有的自适应结构索引的调整粒度是XML元素节点,调整过程往往需要遍历整个文档.而AS-Index是基于F&B-Index节点的增量调整,其过程是局部的,高效的,并且能够支持复杂分支查询的调整.实验结果表明,AS-Index在查询和调整性能上优于现有的XML结构索引.同时,相比于现有的自适应结构索引,AS-Index针对大规模文档具有更加优良的可扩展性.
제출료일충신적자괄응결구색인:AS-Index(adaptive structural index),능구극복현유정태색인화자괄응색인적결함,구비고효적사순화조정성능.AS-Index건립재F&B-Index적기출지상,기색인결구포괄F&B-Index,Query- Table화Part-Table.Query-Table능구기록빈번사순,피면료사순과정중적용여조작.병차,재Query-Table적기출상제출료자저향상적사순처리과정,능구충분이용현유적빈번사순고효지회답비빈번사순.Part-Table용우우화포함조선후예변적사순,진일보제고료사순성능.현유적자괄응결구색인적조정립도시XML원소절점,조정과정왕왕수요편력정개문당.이AS-Index시기우F&B-Index절점적증량조정,기과정시국부적,고효적,병차능구지지복잡분지사순적조정.실험결과표명,AS-Index재사순화조정성능상우우현유적XML결구색인.동시,상비우현유적자괄응결구색인,AS-Index침대대규모문당구유경가우량적가확전성.
This paper proposes an adaptive structural index: AS-Index (adaptive structural index), which can avoid the problem of the existing indexes. AS-Index is based on F&B-Index. It consists of F&B-Index, Query-Table and Part-Table. Frequent queries are kept in Query-Table avoiding redundant operations in query processing. Based on Query-Table an efficient bottom-up query processing is also proposed for answering infrequent queries using the frequent queries in Query-Table. Part-Table is used for optimizing the queries with descendant edges. The existing adaptive structural indexes need to traverse the whole document for adaptation, and their adaptation granularity is XML element node. For AS-Index, the adaptation granularity is F&B-Index node which includes a set of XML element nodes, and its adaptation is an efficient and incremental process that supports branch queries. The experimental results demonstrate that this index significantly outperforms the previous structural indexes in terms of query processing and adaptation efficiencies. For large XML documents, compared with the existing adaptive structural indexes, AS-Index is more scalable