Computational Algorithms for Deep and Shallow Word Search in Sanskrit with Case-inflected Forms of Feminine and Neuter Nouns

Effective and comprehensive word-search in Sanskrit E-text is a non-trivial problem in natural language processing that has been solved in this work for the first time in the literature. The problem is also one of great benefit to a variety of users. The complexity of the problem stems from the fact that words in Sanskrit can get completely metamorphosed on account of the phenomena of euphonic conjunctions and case-inflected forms. This work deals with the case-inflected forms of feminine and neuter gender nouns in particular, and their bearing on the search problem. The novel computational model presented here for generating case-inflected forms of Sanskrit nouns involves a categorization of word forms and a development of new formulae to achieve the generation of case-inflected forms. Pre-processing to improve the performance of the algorithm based on these formulae has also been developed. This model has been extended to include processing of euphonic conjunctions to enable a more elaborate search and to thereby produce meaningful search results. The dual paradigms of deep/slow and shallow/fast searching have been handled in this work and the efficiency of the fast search algorithm developed is shown to have an increased efficiency of 77-80%. The work also discusses the trade-off between search accuracy and speed in the context of word-search in Sanskrit E-text.