哈尔滨工程大学学报
哈爾濱工程大學學報
합이빈공정대학학보
JOURNAL OF HARBIN ENGINEERING UNIVERSITY
2014年
11期
1390-1396
,共7页
董宇欣%王莹洁%宁鹏飞%张耀元
董宇訢%王瑩潔%寧鵬飛%張耀元
동우흔%왕형길%저붕비%장요원
搜索引擎%社会网络%不确定图%SimRank%相似度%抽样策略
搜索引擎%社會網絡%不確定圖%SimRank%相似度%抽樣策略
수색인경%사회망락%불학정도%SimRank%상사도%추양책략
search engine%social network%uncertain graph%SimRank%similarity%sampling strategy
针对以往的搜索引擎日志分析都主要集中在用户行为分析、查询推荐及搜索引擎评价等方面,采用社会网络分析法对搜索引擎进行日志分析。以不确定图的方式逻辑表示搜索引擎的日志中查询词和网页的链接关系,通过基于不确定图的SimRank算法,计算查询词与网页的相似度,最终以相似度和查询词的加权方式建立网页描述库。针对概率抽样的3点基本要求,提出一种渐进式的抽样策略,从而保证采用抽样技术对于不确定图中SimRank值计算的准确性。实验表明该算法具有较好的准确率和可行性。
針對以往的搜索引擎日誌分析都主要集中在用戶行為分析、查詢推薦及搜索引擎評價等方麵,採用社會網絡分析法對搜索引擎進行日誌分析。以不確定圖的方式邏輯錶示搜索引擎的日誌中查詢詞和網頁的鏈接關繫,通過基于不確定圖的SimRank算法,計算查詢詞與網頁的相似度,最終以相似度和查詢詞的加權方式建立網頁描述庫。針對概率抽樣的3點基本要求,提齣一種漸進式的抽樣策略,從而保證採用抽樣技術對于不確定圖中SimRank值計算的準確性。實驗錶明該算法具有較好的準確率和可行性。
침대이왕적수색인경일지분석도주요집중재용호행위분석、사순추천급수색인경평개등방면,채용사회망락분석법대수색인경진행일지분석。이불학정도적방식라집표시수색인경적일지중사순사화망혈적련접관계,통과기우불학정도적SimRank산법,계산사순사여망혈적상사도,최종이상사도화사순사적가권방식건립망혈묘술고。침대개솔추양적3점기본요구,제출일충점진식적추양책략,종이보증채용추양기술대우불학정도중SimRank치계산적준학성。실험표명해산법구유교호적준학솔화가행성。
In this paper, the social network analysis ( SNA) method is adopted to analyze search engine logs ac-cording to previous search engine logs analysis that mainly focuses on user behavior, query recommendation and search engine evaluation.The objective of SNA is to exam the structure of relationships between social entities.This method uses an uncertain graph to logically indicate the link relationship between the query term and the webpage of search engine log.It also computes the similarities between query term and webpage based on the SimRank algo-rithm of an uncertain graph.As well as, establishes a webpage description library by similarity and query term in a weighted way.In addition, this paper proposes a progressive sampling strategy to ensure the calculation accuracy of the SimRank value in an uncertain graph.Experiments show that the algorithm has good accuracy and feasibility.