Introducing the LCC Metaphor Datasets

Introducing the LCC Metaphor Datasets Michael Mohler, Mary Brunson, Bryan Rink, Marc Tomlinson Language Computer Corporation Richardson, Texas, USA michael,mary,bryan,marc@languagecomputer.com Abstract In this work, we present the Language Computer Corporation (LCC) annotated metaphor datasets, which represent the largest and most comprehensive resource for metaphor research to date. These datasets were produced over the course of three years by a staff of nine annotators working in four languages (English, Spanish, Russian, and Farsi). As part of these datasets, we provide (1) metaphoricity ratings for within-sentence word pairs on a four-point scale, (2) scored links to our repository of 114 source concept domains and 32 target concept domains, and (3) ratings for the affective polarity and intensity of each pair. Altogether, we provide 188,741 annotations in English (for 80,100 pairs), 159,915 annotations in Spanish (for 63,188 pairs), 99,740 annotations in Russian (for 44,632 pairs), and 137,186 annotations in Farsi (for 57,239 pairs). In addition, we are providing a large set of likely metaphors which have been indepen- dently extracted by our two state-of-the-art metaphor detection systems but which have not been analyzed by our team of annotators. Keywords: Metaphor, Dataset, Metaphoricity, Conceptual Metaphor, Affect 1. Introduction Metaphor is everywhere in human language. It adorns our poetry, adds clarity to our science and philosophy, and serves as a fountain of language for new and complex ideas. For this reason, it is crucial that natural language pro- cessing (NLP) technologies be capable of correctly iden- tifying and interpreting metaphor. Indeed, metaphor has been found to confound most language processing tools across a wide variety of applications – including textual entailment, text summarization, word sense disambigua- tion, semantic textual similarity, question answering, and event extraction. Unfortunately, research in metaphor to date has largely been hampered by the lack of size, qual- ity, and diversity in the metaphor datasets available to re- searchers (Shutova, 2015). In this work, we present for the first time Language Computer Corporation’s web-scale metaphor datasets, which have been produced by a team of nine annotators working in four languages (American En- glish, Mexican Spanish, Russian, and Farsi) over a three- year period. As a foundation for our metaphor annotation effort, we begin with George Lakoff’s Contemporary Theory of Metaphor (Lakoff and Johnson, 1980; Lakoff, 1993), which popularized the idea of a conceptual metaphor mapping. Within the cognitive framework of a given conceptual mapping, terms pertaining to one concept or domain (the source) can be used figuratively to express some aspect of another concept or domain (the target). For example, the conceptual metaphor “Life is a Journey” indicates a medium within which the target concept “life” may be more easily discussed and understood. This particular mapping allows us to speak of one being stuck in a “dead-end” job, of a crucial decision as being a “fork in the road”, or of someone’s life “taking a wrong turn”. From this idea of the conceptual metaphor, we define our basic unit of annotation: a pair of syntactically-related can- didate terms within a sentence which …

Content truncated to 500 words