Metaphor Identification in Large Texts Corpora

Yair Neuman,Shlomo Argamon,Ophir Frieder,Mark Last,Dan Assaf,Newton Howard,Yohai Cohen,Eduardo G Altmann

doi:10.1371/journal.pone.0062343

Abstract

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.

Highlights

IntroductionThe use of the verb ‘‘fell’’ in the sentence ‘‘The coin fell into the sewer’’ is different from its use in the sentence ‘‘He fell in love’’
Human language comprises figurative and non-figurative dimensions
To the best of our knowledge, the current paper presents the most comprehensive study of metaphor identification in terms of scope of metaphorical phrases and annotated corpora size

Summary

Introduction

The use of the verb ‘‘fell’’ in the sentence ‘‘The coin fell into the sewer’’ is different from its use in the sentence ‘‘He fell in love’’. According to the Oxford English Dictionary, the verb is used in its first sense of ‘‘move downward’’. In the second sentence it is used in its extended metaphorical sense: ‘‘pass into a specified state’’. Differentiating figurative and non-figurative language-use may be highly important for a variety of applications that are based on natural language understanding. Giving a robot the order: ‘‘Give me the bottle’’ is totally different from giving the order: ‘‘Give me a break’’. The robot must have a natural language understanding module that can differentiate between figurative and non-figurative language

Objectives

Results

Conclusion