CAJ | 학술논문

为解决搜索引擎ASPSeek在大规模数据下检索效率低下、占用空间大以及不利于更新等问题,提出了一种分块式存储的倒排索引组织技术,并对基于外存的B+树索引和线性散列索引的性能进行了比较测试研究。测试结果表明,查询每万条数据耗时线性散列为B+树索引快57.40％,插入每万条数据耗时线性散列为B+树索引的2.44倍,删除每万条数据耗时线性散列为B+树索引的83.52％,线性散列索引文件大小为B+树索引文件大小的109.56％。由测试结果可知,B+树索引具有较快的索引构建和更新速度,而线性散列索引则具有较高的磁盘空间占用率和较好的查询性能。
위해결수색인경ASPSeek재대규모수거하검색효솔저하、점용공간대이급불리우경신등문제,제출료일충분괴식존저적도배색인조직기술,병대기우외존적B+수색인화선성산렬색인적성능진행료비교측시연구。측시결과표명,사순매만조수거모시선성산렬위B+수색인쾌57.40％,삽입매만조수거모시선성산렬위B+수색인적2.44배,산제매만조수거모시선성산렬위B+수색인적83.52％,선성산렬색인문건대소위B+수색인문건대소적109.56％。유측시결과가지,B+수색인구유교쾌적색인구건화경신속도,이선성산렬색인칙구유교고적자반공간점용솔화교호적사순성능。
To solve the problem that in large-scale data condition the ASPSeek search engine retrievals inefficiently,has large disk space occupancy and can’t be conducive to update,propose an inverted index-organized technique based on block storage,and make a per-formance comparison research test between external memory based B+tree index and linear hash index.Test results show that,for queries per million data-consuming linear hashing to B+tree index is 57.40%,for inserting per million data-consuming linear hash is 2.44 times to B+tree index,for deleting every million data-consuming linear hash to B+tree index is 83.52%,linear hash index file size is 109.56% of B+tree index file size.According to the test results,B+tree index has the faster index building and updating speed,while linear hash index has the higher disk space occupancy rates and better query performance.