Who are the American Vegans related to Brad Pitt?

Abstract

In this demo, we present Entity Relatedness Graph (EnRG), a focused related entities explorer, which provides the users with a dynamic set of filters and facets. It gives a ranked lists of related entities to a given entity, and clusters them using the different filters. For instance, using EnRG, one can easily find the American vegans related to Brad Pitt or Irish universities related to Semantic Web. Moreover, EnRG helps a user in discovering the provenance for implicit relations between two entities. EnRG uses distributional semantics to obtain the relatedness scores between two entities.

Extracting entity-specific substructures for RDF graph embeddings

“…Semantic Similarity and Relatedness: Semantic similarity and relatedness between two entities have been relatively well explored [1,6,8,14]. Searching for similar or related entities given a search entity is a common task in information retrieval.…”

“…However, before developing such functionality, it is important to define the notion of entity similarity and the set of attributes that will be used for its computation. Semantic similarity and relatedness are often used interchangeably in literature [1,14], where similarity between two entities is computed based on common paths between them. This definition allows computation of similarity between any two entities, including entities of different types.…”

Non-Orthogonal Explicit Semantic Analysis

“…It computes the frequency of a concept appearing as hyperlink in the Wikipedia articles. To obtain the DiSER based relatedness scores between Wikipedia concepts, we use Entity Relatedness Graph (EnRG) 4 (Aggarwal et al, 2015), which is a focused related entities explorer based on DiSER scores.…”

Semantic similarity analysis and application in knowledge graphs

“…Firstly, semantic contents of entities such as textual descriptions and semantic categories are represented in BOW or VSM to compute entity-entity similarity based on: (1) dot or cosine similarity of entity description or category vectors (Cucerzan, 2007); (2) topical coherence between entities using overlap of weighted keyphrases (Hoffart et al, 2012b) and topic models (Piccinno and Ferragina, 2014a); and (3) semantic similarity of entity category hierarchies (Shen et al, 2012a). Secondly, from entity annotated corpora, entity co-occurrence (Nunes et al, 2013) and entity distribution (Aggarwal et al, 2015;Shen et al, 2012b) are used to compute entity-entity relatedness based on the application of distributional hypothesis (Turney et al, 2010) which assumes that entities occur in similar contexts are semantically related. Finally, apart from semantic content analysis and distributional analysis, graph analysis is also very effective in measuring entity connectivity in order to compute entity-entity relatedness, given that entities are connected to each other in KGs.…”

“…Comparing to entity embedding (Zwicklbauer et al, 2016;Fang et al, 2016) based on entityentity co-occurrence (Nunes et al, 2013) and entity distribution (Aggarwal et al, 2015;Shen et al, 2012b), category embedding is independent of annotated data and can have more training data by collecting multiple entities which share the same category. In KGs, an entity usually has multiple categories from general to specific describing different aspects of entity.…”