A Generalisation of Lexical Functions for Composition in Distributional Semantics

Abstract

Over the last two decades, numerous algorithms have been developed that successfully capture something of the semantics of single words by looking at their distribution in text and comparing these distributions in a vector space model. However, it is not straightforward to construct meaning representations beyond the level of individual words -i.e. the combination of words into larger units -using distributional methods. Our contribution is twofold. First of all, we carry out a largescale evaluation, comparing different composition methods within the distributional framework for the cases of both adjectivenoun and noun-noun composition, making use of a newly developed dataset. Secondly, we propose a novel method for composition, which generalises the approach by Baroni and Zamparelli (2010). The performance of our novel method is also evaluated on our new dataset and proves competitive with the best methods.

Integrating Type Theory and Distributional Semantics: A Case Study on Adjective–Noun Compositions

“…In this section, we provide results for a pilot study as to whether the two distributional approaches described earlier reflect a semantic shift in co-composition for adjectives and nouns and can offer something like a justification, in terms of related words, of an expression's use in context. Our evaluation used a list of English adjective-noun combinations drawn from Wiktionary, extracted by the method discussed in Bride, Van de Cruys, and Asher (2015). We added to this list adjective-noun combinations that we thought would exhibit more interesting co-compositional interaction, to achieve a list of 246 adjective-noun pairs in total (see Appendix A).…”

Learning Compositionality Functions on Word Embeddings for Modelling Attribute Meaning in Adjective-Noun Phrases

“…More precisely, we investigate the hypothesis that a full additive model captures a generalized compositional process in the semantics of attributedenoting adjective-noun phrases rather than the lexical meaning of individual attributes (cf. Bride et al (2015)). …”

“…Widdows (2008), Grefenstette and Sadrzadeh (2011), Grefenstette et al (2014)). Bride et al (2015) observe that such composition operators, by being trained on empirical corpus data, can either be tailored to specific lexical types (i.e., individual composition functions for each adjective in the corpus), or designed to capture general compositional processes in syntactic configurations (i.e., a single lexical function for all adjective-noun phrases). In line with these authors, we aim at learning a lexical function which captures attribute meaning in the compositional semantics of adjective-noun phrases, while generalizing over individual attributes.…”

“…As a weighted multiplicative model, we use multiplication of adjective and noun representations with a learned thirdorder tensor T , following Bride et al (2015):…”

“…(iii) We show that f captures generalized attribute meaning (cf. Bride et al (2015)) that abstracts from individual attributes. Thus, after fitting the compositionality function, our model bears the potential of being applied to various application scenarios (e.g., aspect-based sentiment analysis) involving diverse attribute inventories.…”