Entity Linking meets Word Sense Disambiguation: a Unified Approach

Abstract

Entity Linking (EL) and Word Sense Disambiguation (WSD) both address the lexical ambiguity of language. But while the two tasks are pretty similar, they differ in a fundamental respect: in EL the textual mention can be linked to a named entity which may or may not contain the exact mention, while in WSD there is a perfect match between the word form (better, its lemma) and a suitable word sense. In this paper we present Babelfy, a unified graph-based approach to EL and WSD based on a loose identification of candidate meanings coupled with a densest subgraph heuristic which selects high-coherence semantic interpretations. Our experiments show state-of-the-art performances on both tasks on 6 different datasets, including a multilingual setting. Babelfy is online at http://babelfy.org

Automatic understanding of sketch maps using context-aware classification

“…Knowledge is often stored in form of ontologies or similar graph-based structures. This typically leads to a local context that is also defined within graphs (Fenza et al, 2009;Moro et al, 2014).…”

From senses to texts: An all-in-one graph-based approach for measuring semantic similarity

“…PageRank represents the web as a graph and estimates the importance of a web page on the basis of the structural properties of the graph. The algorithm has been successfully used in various fields including NLP where it has found numerous applications: sentiment polarity detection [121], Word Sense Disambiguation [122][123][124][125], semantic similarity [93,94,89], keyword extraction [126], and lexical resource alignment [127,20].…”

Recommendation of microblog users based on hierarchical interest profiles

“…-Babelfy Babelfy (Moro et al 2014) is a graph-based word sense disambiguation (WSD) algorithm based on a loose identification of candidate meanings coupled with a densest sub-graph heuristic which selects highcoherence semantic interpretations. Babelfy disambiguates all nominal and named entity mentions occurring within a text, using the BabelNet semantic network (Navigli et al 2012), a very large multilingual knowledge base obtained from the automatic integration of Wikipedia and WordNet.…”

“…However, as will be shown later in this paper, 82 % of users have at least one topical friend among their friends, and the average number is 12.7. Since, in general, multiple pages may correspond to a user name, we perform joint name resolution and disambiguation using a state-of-the-art online word sense disambiguation algorithm, Babelfy (Moro et al 2014), which disambiguates a textual input against WiBi categories. We use Babelfy as follows: for any user and screen-name, e.g., @britneyspears, we first retrieve from the corresponding Twitter profile the field's name, line À status, and location, e.g., ''Britney Spears'', ''It's Britney...'', ''Los Angeles, CA''.…”

“…3 that in some cases Babelfy returns more than one sense, especially if the senses are closely related, such as Tate and Tate_Britain. Babelfy is a state-of-the-art sense disambiguation algorithm which has already been evaluated in Moro et al (2014). However, we performed a manual evaluation of the name disambiguation task on 400 names for which a sense was retrieved by Babelfy, and also on 400 names from which no senses were retrieved, achieving a 0.81 % F-measure.…”