Processing of the language for extraction of semantic is a challenge in the field of artificial intelligence. Work is being carried out in most of the Indian languages, and Sanskrit being the ancient of all, has a great potential in the extraction process. As given in the grammar of Sanskrit language, its case endings are strong identifiers of the respective word in the sentence. These features are used for identifying the thematic role of a word in a sentence. The problem lies in developing a system with these capabilities. Generally a complete database of all such suffixes is maintained, and then these databases are mapped with each word to identify it as a noun, pronoun, verb and adjective. However, search time in these is very large. This paper presents the design of the system which processes the suffix with the help of the type of net, similar to the transition net for identification purpose. Such extraction features help in representation of knowledge using Sanskrit language.

Knowledge representation and information retrieval are important aspects of any intelligent system. Efficiency of knowledge representation depends on the language in which the information is fed in the system and the capability of that language towards semantic extraction. English language has been analyzed and processed by researchers, and work is being carried out for Indian languages at various research centers like IIIT (Hyderabad), CDAC, JNU (Delhi), etc. Tools like semantic net, conceptual dependencies, frames, etc., are used for knowledge representation with English being the input language. No such tools are available for the other languages; they need to be developed. An effort in this direction is morphological analysis of Sanskrit with respect to linguistic model (Nilson, 2002). This research work uses Sanskrit language for knowledge representation, as it has an excellent grammatical structure. It has also been shown that Panini Grammar Framework (PGF) can be used to develop a suitable computational grammar for free order languages and can successfully be applied to Indian languages (Akshar and Rajeev, 1993). The relation of Panini grammar with order free language and Context Free Grammar (CFG) has also been depicted in the work while establishing the relation between PGF and Western computational framework (Akshar et al., 1995). Sanskrit language has been compared to set theory with rules and meta rules in if-then-else form which shows that the work of Panini in 500 BC was undoubtedly marvelous (Narsingh Rao, 2005). In this work, the use of PGF for knowledge representation is emphasized, and a method for extracting the suffix using transition network is described. The entire problem can be stated as follows:

Given a sentence S in Sanskrit, identify the word W, extract the suffix Sx, and using the case ending analysis, identify its role in the sentence as an agent, object, recipient, instrument, etc. Vibhakti-karka mapping is used to identify the vibhakti and hence karka which gives its role as agent, object, etc. The outline of the algorithm is as follows:

Each of these suffixes in Sanskrit identifies not only the word as a noun, adjective, etc., but also its thematic role in the sentence. In the previous system (Smita and Jyoti, 2007a), a complete search of the database was recommended, but it can be also achieved through the transition network thereby reducing the search time and maintenance of the complete database, as shown in the following section.

Computer Sciences Journal, Transition Networks, Artificial Intelligence, Sanskrit Language, Panini Grammar Framework, Context Free Grammar, Western Computational Framework, Language Processing Systems, Subtransition Networks, Semantic Extractions, Intelligent Systems.