Distributional semantics - Revolvy

Distributional similarity is at best an approximation to semantic similarity.

01/01/1973 · Distributional semantics is a ..

To evaluate the context-weighted similarity measure I compare ranked similarity lists againstgold-standard resources using precision and recall-based measures from Information Retrieval,since the alternative, application-based evaluation, can often be influenced by distributionalas well as semantic similarity.

From Distributional to Semantic Similarity

This talk will describe a complete mathematical framework for deriving distributed representations compositionally using Combinatory Categorial Grammar (CCG). The tensor-based framework extends that of Coecke et al., which was previously applied to pregroup grammars, and is based on the observation that tensors are functions (multi-linear maps) and hence can be manipulated by the combinators of CCG, including type-raising and composition. The existence of robust, broad-coverage CCG parsers opens up the possibility of applying the tensor-based framework to naturally occurring text.

Many semantic generalizations have been amply investigated within thefield of formal semantics, such as the mass-count noun distinction, orthe telicity of verb phrases, inter alia. However, few studies havelooked at these phenomena from an empirical perspective. In this talk,we explore whether empirical data coming from distributionalsemantics is able to support semantic generalizations that are knownfrom formal semantic theories. Secondly, we explore whetherdistributional methods are able to model certain theoretical conceptsknown from formal semantics (such as coercion) in a fully automaticway.


Distributional Semantics: Word Association and Similarity

Semantic knowledge acquisition from parallel corpora relies ondistributional and translational regularities observed in thedata. These serve to cluster word usages into senses and to identifysets of semantically related words and larger text segments. Theresulting synonym and paraphrase sets offer a generalisation acrosssimilar meanings which can reduce sparseness and improve evaluation intranslation applications. In this talk I will look into the relationsencoded in semantic representations commonly used in multilingualtasks. I will demonstrate that meaning distinctions present inresources derived from parallel data might lead to erroneous judgmentsof semantic relatedness. Furthermore, I will explore the correlationof semantic adequacy judgments based on source and translationcontext. I will end with an account of the factors defining semanticrelatedness and substitution in a multilingual setting.

Characterising measures of lexical distributional similarity

The concept of (discrete) types and type hierarchies is amainstay of formal semantics. In contrast, distributionalsemantics is generally based on a notion of (graded)similarity and, correspondingly, (graded) typicality.

distsim.4 - Distributional Similarity A common method …

These representations are objective or empiricist, based on a generalisation of distributional learning, and are capable of representing all regular languages, some but not all context-free languages and some non-context-free languages.

Introduction to Distributional Semantics - Data & …

This workshop aims to elicit interaction between the fields offormal semantics and distributional semantics. Formal semantics is thediscipline that tries to produce a very precise theory of meaning byrepresenting natural language within a logical framework. Theprinciple of compositionality has been explored in depth within thedomain of formal semantics. However, formal semanticists mostly focuson functional elements, such as determiners, quantification, modality,and tense morphemes, . The standard representation oflexical items within a formal semantic framework has been somewhatneglected: the meaning of a content word is usually represented by itsdenotation, which is difficult to implement within a computationalframework. Distributional semantics, on the other hand, is verysuccessful at inducing the meaning of individual content words, butless so with regard to function words and compositionality. In thisrespect, the frameworks of formal semantics and distributionalsemantics are complementary. Still, there has not been a lot ofinteraction between both communities throughout the years. Thisworkshop brings together researchers from both fields of formal anddistributional semantics, in order to promote interaction between bothcommunities.