Linked Data Entity Resolution System Enhanced by Configuration Learning Algorithm

Abstract

SUMMARY Linked data entity resolution is the detection of instances that reside in different repositories but co-describe the same topic. The quality of the resolution result depends on the appropriateness of the configuration, including the selected matching properties and the similarity measures. Because such configuration details are currently set differently across domains and repositories, a general resolution approach for every repository is necessary. In this paper, we present cLink, a system that can perform entity resolution on any input effectively by using a learning algorithm to find the optimal configuration. Experiments show that cLink achieves high performance even when being given only a small amount of training data. cLink also outperforms recent systems, including the ones that use the supervised learning approach.

Data linking over RDF knowledge graphs: A survey

“…All the candidates pairs belonging to S are labeled positive (ie, ( s , t )∈ S ). The other candidates pairs that include the first element ( s ) of an existing positive labeled pair are labeled negative (ie, ( s , z ) where z ≠ t is labeled negative) . Finally, the candidates pairs are divided into training and validation sets and they are provided to the fourth configuration module.…”

“…cLink is a supervised approach, which is implemented as a part of ScSLINT. It uses a heuristic search method permitting to learn an optimal matching configuration (ie, combination of similarity functions as well as other settings of the IM process) defined in ScSLINT.…”

“…In contrast, the class‐disambiguation applied in SERIMI seems to be effective in refining candidates especially for the dataset‐pairs, Sider‐Diseasome, and Sider‐Drugbank, which were problematic for the alternative approaches. We note that the results for Paris are reported from cLink paper …”

Genetic-Fuzzy Programming Based Linkage Rule Miner (GFPLR-Miner) for Entity Linking in Semantic Web

“…ActiveAtlas,TAILOR,MARLIN,andGenLinkarelinkingsystemsbasedonsupervised learningmodels.InActiveAtlas (Tejada,Knoblock,&Minton,2001)andTAILOR (Elfeky& Verykios,2002)authorshavedesignedadecisiontreebasedclassifiersystemtoidentifymatches andnon-matchestolinkentities.ActiveAtlasusesaC4.5algorithmtobuildadecisiontreewhile TAILORusestheID3algorithm.MARLIN(MultiplyAdaptiveRecordLinkagewithInduction) (Bilenko&Mooney,2003)learnsstringsimilaritymeasuretocalculatethesimilaritybetween entitypropertiesandusesasupportvectormachine (Cortes&Vapnik,1995)tocombinelearned similaritymeasurestodesignaclassifiermodel.Insupervisedlearning,basedapproachesmost closedtoourworkareGenLink (RIsele&Bizer,2011),EAGLE(Ngomoetal.,2012,cLink and ScLink (Nguyen & Ichise, 2016a, 2016b. The GenLink and EAGLE systems use GP to learnLRsfromtheexistingsetoflinks.IncLinkandScLinkauthorshaveusedheuristicsto optimallylearnconfigurationstoperformlinking.…”