Implemented Matchers Sample Clauses

Implemented Matchers. In porting matchers from AgreementMaker, a paramount idea was that, when- ever possible, we would eliminate the need for an all-against-all comparison and instead convert them into Primary Matchers whenever possible. We imple- mented four matchers: the Lexical Matcher, the Mediating Matcher, the Word Matcher and the Parametric String Matcher. The first three were implemented as Primary Matchers and the last one was implemented as a Secondary Matcher. Lexical Matcher is one of the simplest and most efficient matching algorithms, which looks for literal name matches in the Lexicons of the input ontologies. It is derived from the Lexical Synonym Weighted Matcher of AgreementMaker, but uses a more streamlined weight system. For two input ontologies source and target, the Lexical Matcher algorithm is as follows: set A = empty alignment set list = names(source) for each name in list if target contains name set sourceList = sourceClasses(name) set targetList = targetClasses(name) for each sourceClass in sourceList set weightSource = weight(name,sourceClass) for each targetClass in targetList set sim = weightSource * weight(name,targetClass) add (sourceClass,targetClass,sim) to A end Note that, regardless of the order in which the input ontologies are given, the Lexical Matcher always iterates through the Lexicon of the ontology that has the smallest number of names, so as to minimize the number of iterations. Mediating Matcher is also a simple and efficient matching algorithm based on literal name matches, but it uses a third (external) ontology as a mediator between the input ontologies. It uses the Lexical Matcher to compute “bridge” alignments between the input ontologies and the mediating ontology, then cross- searches the bridge alignments to map the input ontologies. For two input on- tologies source and target and a mediating ontology med, the Mediating Matcher algorithm is as follows: set A = empty alignment compute Bs = LexicalMatcher(med,source) compute Bt = LexicalMatcher(med,target) for each medClass in Bs if Bt contains medClass for each sourceClass in Bs(medClass) set simSource = similarity(medClass,sourceClass) for each targetClass in Bt(medClass) set sim = simSource * similarity(medClass,targetClass) add (sourceClass,targetClass,sim) to A end Word Matcher is a word-based string similarity algorithm that measures the similarity between two classes through a weighted Jaccard index between the words present in their names. It is derived from the Vector-ba...