Формування та репрезентація складних слів для комп’ютерної обробки: підхід універсального семантичного представлення з Панінійської перспективи

Abstract

Understanding the semantic structure of compound formations is a central concern in both theoretical and computational linguistics. The Universal Semantic Representation (USR) provides a computationally tractable framework for capturing meaning in a structured, language-independent manner, drawing inspiration from the Indian Grammatical Tradition (IGT). This study applies Pāṇinian principles of compounding to the USR framework in order to analyse compound structures from a Pāṇinian perspective. While principles such as ākāṅkṣā (expectancy) and yogyatā (semantic congruity) are foundational to semantic interpretation, the study particularly demonstrates how specific Pāṇinian constraints, including “saviśeṣaṇānāṃ vṛttir na” and “vṛttasya vā viśeṣaṇayogo na” help resolve modifier-scope ambiguities in automated systems and semantic parsing. The paper further examines the internal semantic organisation of compound expressions and the hierarchical relations that emerge in nested and complex compound constructions. Incorporating these insights into the USR framework demonstrates how compound semantics, dependency relations, and contextual interpretation can be systematically represented across languages for computational processing. Examples drawn from Hindi geography textbooks illustrate the practical applicability of the proposed analysis and underscore the relevance of Pāṇinian grammar for advancing computational approaches to semantic representation and Natural Language Generation (NLG). The study analyses 1,504 sentences from Hindi geography textbooks and observes that compound density reaches approximately 52 % in technical chapters, thereby necessitating a systematic logic for representing nested compound structures and preserving semantic integrity. The findings suggest that the integration of Pāṇinian grammatical insights with USR contributes significantly to Natural Language Processing (NLP) models for Indian languages by improving semantic precision, interpretive consistency, and computational representation of complex linguistic structures. The study highlights the continuing relevance of traditional Indian grammatical theories in contemporary computational linguistics and knowledge representation research.