DWRank: Learning concept ranking for ontology search

Abstract

With the recent growth of Linked Data on the Web there is an increased need for knowledge engineers to find ontologies to describe their data. Only limited work exists that addresses the problem of searching and ranking ontologies based on a given query term. In this paper we introduce DWRank, a two-staged bi-directional graph walk ranking algorithm for concepts in ontologies. DWRank characterises two features of a concept in an ontology to determine its rank in a corpus, the centrality of the concept to the ontology within which it is defined (HubScore) and the authoritativeness of the ontology in which it is defined (AuthorityScore). DWRank then uses a Learning to Rank approach to learn the feature weights for the two aforementioned ranking strategies. We compare DWRank with state-of-the-art ontology ranking models and traditional information retrieval algorithms. This evaluation shows that DWRank significantly outperforms the best ranking models on a benchmark ontology collection for the majority of the sample queries defined in the benchmark. In addition, we compare the effectiveness of the HubScore part of our algorithm with the state-of-the-art ranking model to determine a concept centrality and show the improved performance of DWRank in this aspect. Finally, we evaluate the effectiveness of the design decisions made for the AuthorityScore method in DWRank to find missing inter-ontology links and present a graph-based analysis of the ontology corpus that shows the increased connectivity of the ontology corpus after extraction of the implicit inter-ontology links.

A Taxonomy of Semantic Web Data Retrieval Techniques

“…In these approaches, different graph/ontology features are selected (or computed) and on the basis of these features a ranking model is learnt and then the learned model is used to produce the ranking for search results [8].…”

Popularity-Driven Ontology Ranking Using Qualitative Features

“…Approaches to ontology ranking adopt conventional ranking techniques and models from information retrieval, which can be categorized as follows [17]: relevance ranking models aim to rank an ontology o from a repository R based on their relevance to a query q, i.e., in the form of Φ(q, o) or Φ(q, o, R). These include wellknown approaches (e.g., TF-IDF [26], BM25 [24]) and further ontology-specific approaches such as centrality of matched concepts in the ontology graph [8]. On the other hand, one can find importance ranking models that rank ontologies independently from the query, i.e., in the form of Φ(o) or Φ(o, R).…”

“…The CBRBench ground truth [6] was gathered through human labeling based on how well ontology terms meet their definition in a dictionary, comprising ten queries with a total of 819 relevance judgments. CBRBench was used to learn a ranking model in DWRank [8]. Termpicker [28] proposes a ground truth derived from LOD datasets and a ranking model that relies on popularity features, offering ontology term recommendations upon a query in form of triple patterns.…”

A Framework for Evaluating Snippet Generation for Dataset Search

“…We give particular attention to methods that are capable of generating query-biased snippets for ontology search [3,17,31,5,6]. An ontology schema is often represented as a graph where nodes represent terms and edges represent axioms associating terms [44,17].…”