Sadrzadeh/Grefenstette: "Compositional Distributional Models of Meaning"

Ponentes: Mehrnoosh Sadrzadeh y Edward Grefenstette (Computing Laboratory, Oxford University)

Fecha: 28 de abril de 2011

Lugar de celebración: Sala 2.24, Facultad de Psicología, UNED

Resumen

Mehrnoosh Sadrzadeh: Mathematical Foundations for Compositional Distributional Models of Meaning (10h00 - 12h00) [pdf slides]

This tutorial will introduce a compositional distributional model of meaning and will explain the mathematical concepts behind it. The model, jointly developed in Oxford and Cambridge, provides a compositional theory of meaning that computes meanings of sentences as vectors and is able to measure their similarity. The ingredients include a type-logic for formalizing the grammatical structure (called a pregroup), and a vector space semantics for lexical meaning. I will introduce the key concepts of each setting and go through simple fun examples to illustrate their applicability. I will also work through sample computations which parse the sentence, assign meaning to words, then compose them to obtain meaning of sentences.

Edward Grefenstette: Empirical Validation of Compositional Distributional Models of Meaning (12h30 - 14h00) [pdf slides]

Modelling compositional  meaning for sentences using empirical distributional methods has been a challenge for computational linguists. We implemented the abstract categorical model discussed in the previous talk using data from the BNC and evaluated it. In this talk, we will present the implementation, based on unsupervised learning of matrices for relational words and applying them to the vectors of their arguments. We'll also discuss the results of the evaluation, based on the word disambiguation task developed by Mitchell and Lapata (2008) for intransitive sentences, and on a similar new experiment designed for transitive sentences. We'll show our model to match the results of its competitors in the first experiment, and better them in the second. The general improvement in results with increase in syntactic complexity showcases the compositional power of our model.

Bio

Dr Mehrnoosh Sadrzadeh is an EPSRC Advanced Research Fellow at Oxford University's Computing Laboratory. Her PhD was in logical methods in computer science and in particular algebraic and categorical methods to reason about information and in formalizing grammar of natural languages via syntactic word order and compositional distributional semantics. She has authored and co-authored two book chapters on logical methods in linguistics and in epistemology, as well as published over 25 articles in similar areas in journals and refereed conference and workshop proceedings. Together with Clark and Coecke she developed the first compositional distributional model of meaning.

Edward Grefenstette is an ESPRC DTA funded DPhil candidate at the Oxford University's Computing Laboratory in the fi eld of computational linguistics, supervised by Dr Coecke and Professor Pulman FBA. A physicist by his fi rst degree, he has a Masters in philosophy of mathematics and logic and another in computational linguistics. He has been actively involved in developing and evaluating concrete models for constructing meaning of sentences in the recent categorical setting developed by Clark, Coecke, and Sadrzadeh, with whom he has collaborated in several recent papers.

Lugar de Celebración

Sala 2.24 (segunda planta)
Facultad de Psicología, UNED
c/ Juan del Rosal, 10
28040, Madrid

Horarios

jueves 28 de abril de 2011: 10h00 - 14h00

Materiales

Mathematical Foundations for Compositional Distributional Models of Meaning.

Empirical Validation of Compositional Distributional Models of Meaning.

 
RocketTheme Joomla Templates