Metaphor Identification in Persian: Annotation, Data Analysis, and Reliability Assessment for Compiling a Metaphor Corpus for Persian

This is becoming increasingly necessary for machines to be able to comprehend figurative language as artificial intelligence and natural language processing continue to advance. Metaphors are one of the figurative language forms that are difficult for machines to grasp. Developing metaphor corpora, which will enable machines to be trained using them, is the initial stage in the process of enhancing metaphor comprehension. The Metaphor Identification Procedure Vrije Universiteit, known as MIPVU, is a method that can be used to annotate metaphors. In the present study, MIPVU is evaluated in order to compile a Persian metaphor corpus. A collection of scholarly papers and news articles was gathered and annotated. The reliability of the procedure was subsequently evaluated using the Kappa coefficient and Cochran's Q. According to the results of the investigation, MIPVU can annotate Persian metaphors precisely and reliably (κ=0.964). Consequently, this procedure offers a reliable method for compiling a metaphor corpus.