Table 2.

Summary of IR algorithms

Algorithm	Scoring	Global	WPM	Remarks
tf-idf	Term frequency	No	No	Frequent resources in the collection have a low score. In ontologies, a common term does not necessarily mean less relevant. Frequent terms can be a product of reuse by other ontologies
BM25	Term frequency	Yes	No	Suffers from the same issue has tf-idf, but the cumulative score ranks domain ontologies higher
VSM	Vector similarity	No	No	Uses tf-idf to weight vectors and also considers the tf-idf of the query, aggravating the tf-idf drawback
PageRank	Links between ontologies	Yes	No	Ranks based on popularity, which may lead to popular but less relevant resources, being ranked higher
CMM	Coverage of the set of queries	Yes	Yes	Ontologies with a large number of partial matches will be scored higher than ontologies with few exact matches
SMM	Closeness between ontological resources	Yes	No	Although this algorithm can be useful when considering similarity among the matched resources of two or more query terms of a multi-keyword query, it performs poorly on single-word queries

Note: Scoring summarizes the main scoring method of the algorithm. Global indicates if the score attributed by the algorithm is per resource or per ontology. WPM (weights partial matches) shows if the ontology distinguishes between partial and exact matches.