【摘要】：子圖查詢是圖數(shù)據(jù)庫(kù)研究中的一個(gè)重要問題,許多方法基于"過濾-驗(yàn)證"策略進(jìn)行子圖查詢,算法研究的重點(diǎn)為快速找到有效的特征集.通過對(duì)特征模式在數(shù)據(jù)圖集中的嵌入信息進(jìn)行分析,離線建立基于重疊關(guān)系、鄰接關(guān)系和近鄰關(guān)系的嵌入關(guān)系索引,提出基于嵌入關(guān)系的子圖查詢算法ERSearch.在給定查詢圖后,利用特征共現(xiàn)關(guān)系與特征嵌入關(guān)系聯(lián)合進(jìn)行過濾操作,并將過濾階段的嵌入關(guān)系比對(duì)結(jié)果用于驗(yàn)證過程,提高驗(yàn)證效率.在真實(shí)及模擬數(shù)據(jù)上的實(shí)驗(yàn)表明,通過與PathIndex等方法的對(duì)比,ERSearch算法有效縮減了候選集的規(guī)模,能有效提高過濾與驗(yàn)證階段的執(zhí)行效率.
[Abstract]:Subgraph query is an important problem in the research of graph database. Many methods perform subgraph query based on "filter-verify" strategy. The focus of algorithm research is to find effective feature sets quickly. By analyzing the embedding information of feature pattern in the data set, the embedding relation index based on overlapping relation, adjacency relation and nearest neighbor relation is established offline, and a subgraph query algorithm ERSearch. based on embedded relation is proposed. After the query graph is given, the feature co-occurrence relation and the feature embedding relation are combined to carry out the filtering operation, and the results of the comparison of the embedding relations in the filtering stage are used in the verification process to improve the verification efficiency. Experiments on real and simulated data show that, by comparing with PathIndex and other methods, ERSearch algorithm can effectively reduce the size of candidate sets and improve the efficiency of filtering and verification.
【作者單位】： 中山大學(xué)信息科學(xué)與技術(shù)學(xué)院;吉首大學(xué)軟件服務(wù)外包學(xué)院;廣州中醫(yī)藥大學(xué)醫(yī)學(xué)信息工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金(No.61033010,No.61272065,No.61472453) 廣東省自然科學(xué)基金(No.S2011020001182,No.2014A030309013) 廣東省科技計(jì)劃基金(No.2009B090200450,No.2010A040303004,No.2011B040200007) 廣東省醫(yī)學(xué)科研基金(No.B2014174)
【分類號(hào)】：TP311.13

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王映龍;楊s，

本文編號(hào)：2454565