Neurostimulation for Finding Metaphor in Terminology

The article describes the experiment aimed to automate the metaphor identification procedure in terminology. For a metaphor, we take a term whose contextual specialized meaning is determined based on the meaning realized in daily, concrete, or body-related meaning. For instance, the term computer virus is understood through biological viruses. The task is nontrivial. Metaphor Lab (Amsterdam, Netherlands) has algorithmized the procedure in nonspecialized language. However, the task still lacks automation. Nor has it been applied to specialized language and terminology.We address this task through artificial intelligence. The experiment employs Neurostimulator 5.0, software designed to model a multilayer neural network with a direct signal propagation to forecast a term’s metaphorical status based on linguistic parameters manually assigned to the terms. The training set comprises more than 100 terms from the computer security domain. In addition, the parameters are retrieved from open-access online dictionaries, which provides further vistas for data parsing and full-stock automation of the metaphor identification process.