The Relational Parts of Speech in Text Analysis for Definition Detection, for Romanian Language
Cristian Niculita, Luminiţa Dumitriu · 2019
This article presents a method for detecting definitions in text, as well as identifying their formal parts, using the predictive capacity of certain relational parts of speech, mainly prepositions but also including words from other lexical categories such as adverbs, pronouns and conjunctions. The method is specifically tailored for Romanian language. Based on their predicting capacity, the relational words can be distributed into four categories that influence the way the definitions are detected. Using an annotated corpus of definitions, a series of statistics is extracted and used to formulate measures describing the impact and usage patterns of these relational words.