Use Of pAninI Grammar For Knowledge Representation And Information Retrieval

Abstract

Intelligent System requires ontological representation for learning purpose and collection of such knowledge entities, paves the way for machine learning. Knowledge representations structures stores the semantic content embedded within the text. Efficiency of these structures is measure of their capability to represent the semantic of the text accurately, facilitating the information retrieval process. Apart from these parameters, language in which the document is written ia also important. Syntactic and semantic analysis is crucial aspect of natural language. For fixed order language, various statistical methods exist and abundant corpus is available for processing. Indian languages lack availability of such source. Ancient Indian Language, Sanskrit, is being explored for similar bahaviour. Fascinated by the syntactic structure of the language, and association of syntactic and semantic feature of words and sentence, language is processed for development of the knowledge representation tool. Research on Indian regional language is gearing up, but lack of syntactic and semantic tools decelerate the work. Methods developed for fixed order language are not suitable for Sanskrit language as it is free order word language. Hence, different mechanism is needed for such language. Development od knowledge representation structure for sanskrit language is the aim of the current work. Entire processing is centred around the concept of pAninI grammar, which determines the relation of words in a sentence with the action in form of cases- agent, object etc. This association is captured and stored in the frame called case frame. A rule based system for this work is analysed, a neural network based system is also developed and hybrid model which blends the previous two methods is also discussed. This knowledge structure is a baseline model for semantic representation of the text which identifies six main roles of words with the verb as per pAninI framework. Another important semantic driven technique is newline