National Journal of System and Information Technology

Sohil D. Pandya , Paresh V. Virparia

Abstract

In this era of Knowledge, organizations can gain competitive advantage only by proficient data analysis. This paper emphasizes on application of clustering in context free data cleaning by correcting values of attributes, using various sequence similarity metrics, where reference data set is not available, to improve the quality of data which in turn lead to eminent data analysis. Authors propose an algorithm to examine suitability of value to correct other values of attributes. Various sequence similarity metrics were used, to find distance of two values of attributes, to test the data and generate results. Experimental results show how the approach can effectively clean the data without reference data.

Keywords

Clustering, Context Free Data Cleaning, Sequence Similarity Metrics

Full Text

     

Jigisha K. Patel ¹, Pritesh N. Patel ¹, Paresh V. Virparia ¹

Affiliations
1 Dept. of Computer Science, Sardar Patel University, IN

Abstract

Natural Language Processing provides the facility of operating a system with speech and the system can interact with the user accordingly. Many speech recognition applications also provide the support for regional languages. In this paper, we would like to discuss the work that will add some facility to the state-of-the-art facility of speech recognition of Gujarati language. We have designed and developed a system that allows the users to give commands to their smartphone in Gujarati language for some basic facilities like calling, sending SMS, etc. The vocabulary includes total of 60 words consisting of Gujarati digits, some persons' name to be considered as a contact name and operational commands which yield the overall average recognition accuracy of 82.23%.

Keywords

Speech Recognition, Phoneme, Hidden Markov Model (HMM), Java Grammar, Lexeme.

Full Text