Encyclopedic Knowledge Patterns from Wikipedia Links

Abstract

Abstract.What is the most intuitive way of organizing concepts for describing things? What are the most relevant types of things that people use for describing other things? Wikipedia and Linked Data offer knowledge engineering researchers a chance to empirically identifying invariances in conceptual organization of knowledge i.e. knowledge patterns. In this paper, we present a resource of Encyclopedic Knowledge Patterns that have been discovered by analyizing the Wikipedia page links dataset, describe their evaluation with a user study, and discuss why it enables a number of research directions contributing to the realization of a meaningful Semantic Web.

From hyperlinks to Semantic Web properties using Open Knowledge Extraction

“…For example, a link to "Usenet" in the Wikipedia page of "John McCarthy" suggests a semantic relation between those two entities, which is explained by the sentence: "McCarthy often commented on world affairs on the Usenet forums" 6 . Besides common sense, this hypothesis is also supported by a previous study [33], which describes the extraction of encyclopedic knowledge patterns for DBpedia types, based on links between Wikipedia pages. A user study showed that hyperlinks between Wikipedia pages determine relevant descriptive contexts for DBpedia entities at the type level, which suggests that these links mirror relevant semantic relations between entities.…”

“…We compared Legalo predicate λ for a triple (s, e subj , e obj ) with all λ i created by the users for that triple. Two different similarity measures were computed: a string similarity score based on Jaccard distance measure 33 , and a semantic similarity measure based on the SimLibrary framework [36] 34 . The latter is a semantic similarity score that extends string similarity with measures exploiting external semantic resources such as such as WordNet, MeSH or the Gene Ontology.…”

“…The Stanford CoreNLP framework 35 was used to compute the lemma and POS tag of each term in the phrase. This lemmatisation step was necessary in order to en- 33 Given two strings s 1 and s 2 , where c 1 and c 2 are the two character sets of s 1 and s 2 , the Jaccard distance J(s 1 ,s 2 ) is defined as J sim (s 1 , s 2 ) = J sim (c 1 , c 2 ) = Also for Hypothesis 2, Legalo shows very satisfactory performance. An impressive result is the high average value of the semantic similarity score (0.80) between user created predicates and Legalo generated ones.…”

Mapping Keywords to Linked Data Resources for Automatic Query Expansion

“…One way we foresee for privileging one concept rather than the other in the simultaneous matching is exploiting Encyclopaedic Knowledge patterns (EKPs) [13], to exploit not only available relations between candidate concepts, but also statistical evidence in the data about connections which have been actually used in instance data.…”

Knowledge Pattern Extraction and Their Usage in Exploratory Search

“…For this purpose some work has been done. In [7] we present a method that we have defined for extracting KPs from Wikipedia by analysing links and their interpretation through the DPpedia Ontology. 4 The analysis of results shows a bias due to a large number of untyped entities in DBpedia.…”