Yohai Cohen

Metaphor Identification in Large Texts Corpora

Identifying metaphorical language-use (e.g., sweet child) is one of the challenges facing natural language processing. This paper describes three novel algorithms for automatic metaphor identification. The algorithms are variations of the same core algorithm. We evaluate the algorithms on two corpora of Reuters and the New York Times articles. The paper presents the most comprehensive study of metaphor identification in terms of scope of metaphorical phrases and annotated corpora size. Algorithms’ performance in identifying linguistic phrases as metaphorical or literal has been compared to human judgment. Overall, the algorithms outperform the state-of-the-art algorithm with 71% precision and 27% averaged improvement in prediction over the base-rate of metaphors in the corpus.

How Language Enables Abstraction: A Study in Computational Cultural Psychology

The idea that language mediates our thoughts and enables abstract cognition has been a key idea in socio-cultural psychology. However, it is not clear what mechanisms support this process of abstraction. Peirce argued that one mechanism by which language enables abstract thought is hypostatic abstraction, the process through which a predicate (e.g., dark) turns into an object (e.g., darkness). By using novel computational tools we tested Peirce’s idea. Analysis of the data provides empirical support for Peirce’s mechanism and evidence of the way the use of signs enables abstraction. These conclusions are supported by the in-depth analysis of two case studies concerning the abstraction of sweet and dark. The paper concludes by discussing the findings from a broad and integrative theoretical perspective and by pointing to computational cultural psychology as a promising perspective for addressing long-lasting questions of the field.

Why “dark thoughts” aren't really dark: A novel algorithm for metaphor identification

Distinguishing between literal and metaphorical language is a major challenge facing natural language processing. Heuristically, metaphors can be divided into three general types in which type III metaphors are those involving an adjective-noun relationship (e.g. “dark humor”). This paper describes our approach for automatic identification of type III metaphors. We propose a new algorithm, the Concrete-Category Overlap (CCO) algorithm, that distinguishes between literal and metaphorical use of adjective-noun relationships and evaluate it on two data sets of adjective-noun phrases. Our results point to the superiority of the CCO algorithm to past and contemporary approaches in determining the presence and conceptual significance of metaphors, and provide a better understanding of the conditions under which each algorithm should be applied.

Change in the embedding dimension as an indicator of an approaching transition.

Predicting a transition point in behavioral data should take into account the complexity of the signal being influenced by contextual factors. In this paper, we propose to analyze changes in the embedding dimension as contextual information indicating a proceeding transitive point, called OPtimal Embedding tRANsition Detection (OPERAND). Three texts were processed and translated to time-series of emotional polarity. It was found that changes in the embedding dimension proceeded transition points in the data. These preliminary results encourage further research into changes in the embedding dimension as generic markers of an approaching transition point.

A generic method for measuring the potential number of structure-preserving transformations

In many fields of psychology, it may be interesting to measure the potential number of structure-preserving transformations that exist between succeeding structures. The aim of this article is to present a methodology for measuring the potential number of structure-preserving transformations between succeeding structures and to illustrate the applicability of the methodology through a case study. The article concludes by discussing the lessons and implications of the proposed methodology for microgenetic research.

A cognitively motivated word sense induction algorithm

The way in which word senses are produced and identified is of great interest to cognitive sciences as well as to various applications in natural language processing. In this paper, we present a cognitively inspired algorithm of word sense induction. The algorithm fuses the distributional and perceptual information of words. By drawing on minimal resources - word collocations and their level of concreteness/abstractness - our algorithm automatically produces for each target noun a graph that is an endomap with a maximal number of 50 nodes. This graph represents the major senses associated with the noun. Tested on a word sense disambiguation task and on psychological data, our algorithm gains significant empirical support for its efficiency.