Learning Conceptual Spaces with Disentangled Facets

Proceedings of the 23rd Conference on Computational Natural Language Learning, pages 131–139 Hong Kong, China, November 3-4, 2019. c⃝2019 Association for Computational Linguistics 131 Learning Conceptual Spaces with Disentangled Facets Rana Alshaikh School of CS & Informatics Cardiff University, UK alshaikhr@cardiff.ac.uk Zied Bouraoui CRIL - CNRS & Univ Artois France zied.bouraoui@cril.fr Steven Schockaert School of CS & Informatics Cardiff University, UK schockaerts1@cardiff.ac.uk Abstract Conceptual spaces are geometric represen- tations of meaning that were proposed by G¨ardenfors (2000). They share many similari- ties with the vector space embeddings that are commonly used in natural language process- ing. However, rather than representing entities in a single vector space, conceptual spaces are usually decomposed into several facets, each of which is then modelled as a relatively low- dimensional vector space. Unfortunately, the problem of learning such conceptual spaces has thus far only received limited attention. To address this gap, we analyze how, and to what extent, a given vector space embedding can be decomposed into meaningful facets in an unsupervised fashion. While this problem is highly challenging, we show that useful facets can be discovered by relying on word embed- dings to group semantically related features. 1 Introduction Conceptual spaces (G¨ardenfors, 2000) are vec- tor space models that are aimed at representing the entities of a given kind (e.g. movies), to- gether with their associated properties (e.g. scary) and concepts (e.g. thrillers). As such, they are similar in spirit to the vector space models that have been proposed in information retrieval (Deer- wester et al., 1990) and natural language pro- cessing (Turney and Pantel, 2010; Mikolov et al., 2013), but there are also notable differences. First, in the context of conceptual spaces, an explicit dis- tinction is made between the entities from the do- main of discourse, which are represented as vec- tors, and the corresponding properties and con- cepts, which are represented as regions (e.g. poly- topes) or soft regions (e.g. characterized by a Gaussian). Second, conceptual spaces are organ- ised into a set of facets1, each of which captures 1These facets are often referred to as domains in the con- text of conceptual spaces. However, we will use the term a different aspect of meaning. For instance, in a conceptual space of movies, we may have facets such as genre, language, geographic location, etc. Each facet is associated with its own vector space, which intuitively captures similarity w.r.t. the corresponding facet. For instance, in a concep- tual space of movies, the vector space for the bud- get facet would only capture whether two movies had a similar budget. Most of these facet spaces tend to be low-dimensional (e.g. modelling budget only needs a single dimension). This clearly dif- ferentiates them from traditional semantic spaces, which often have hundreds of dimensions. From an application point-of-view, the separation of vector space models into facets is appealing for several reasons. One key advantage is that it al- lows us to model similarity in a more flexible, and cognitively more plausible way. A related advantage …

Content truncated to 500 words