Đang chuẩn bị liên kết để tải về tài liệu:
Managing and Mining Graph Data part 62

Managing and Mining Graph Data part 62 is a comprehensive survey book in graph data analytics. It contains extensive surveys on important graph topics such as graph languages, indexing, clustering, data generation, pattern mining, classification, keyword search, pattern matching, and privacy. It also studies a number of domain-specific scenarios such as stream mining, web graphs, social networks, chemical and biological data. The chapters are written by leading researchers, and provide a broad perspective of the area. This is the first comprehensive survey book in the emerging topic of graph data processing. . | 000 MANAGING AND MINING GRAPH DATA scribed was developed in which the L2 models are replaced by a ranking perceptron 53 . Specifically N binary onc-vs-rcsl. SVM models are trained which form the set of L1 models. Similar to the cascade SVM method the representation of each compound cs the training set for the L2 models con-si-ts of its dcscrlptor-bpacc based representation tend its output from each of -he N L1 modelSi Hnaily. ft aanking mo del W learned using the ranking perceptron described ln the prevroua section. Since the L2 model ts Sashd on the dcrcriptoa-epacc bated reprssantation and the outputs of the L1 mode-s the siie of W ls N x n N . 5.2 Performance of Target Fishing Strategies Ahi extensive evaiuation of the i tllTcr si . Target Fishing methods was performed recently rS5S winch primarily used the PubChem 39 database -o extract -argen-spectfic dosc-scsponsc confirmatory assays. Specifically the ablhty oh Ihc five mclhoda to identify relevant categories in the top-fc ranked categories was as-ecsad in this work The sesults were analyzed along this titi Cisiion becaule thin clri-cclly corresponds to Sic use case scenario where a uaco may wan to look af top-fc paedheted tasgets for as test compound and fur--her study or analyze th-m Sor i.oxicli.yt paoimscully. off-target effects pathway endysis etc 53 . The comparlsoas utilized paeersion and recall metric ln Sop-fc lirr each of fire five schones. as shown in Figures 19.3a and 19.3b . TCess tigurca rhow the actual precilion and recall values in top-fc by varying fc from me to fifteen. TCcsc pguaes lndical.c l.haf l os i de s t.H y ing one of the correct categories or tar-gell in the top - pccdtefionSi cascade SVM outperforms all the other schemes ln hems of both precision aad recall. However as fc lnce ac a faom one to fifteen lire precision and tccat 1 rceuh.s ht dhes that the best performing scheme ls the SVM Ranking Pcrccpt.ron s i nd tl. outperforms all other schemes for both paecssion as well ae .