A Unique Query Processing Framework using Lexical-Cepstral Feature Extraction based B2DT Classifier in Natural Language Processing

The amount of data produced today is constantly increasing.With the advent of contemporary database tools and rising technology, we can store a lot of data.But, the problem is that a lot of people need to grow more adapted to the user interfaces and technological advancements that process data and show it in the manner desired by the user.It implies that a huge number of individuals require additional database management expertise.Thus, this work intends to implement a technology that will help users get precise data from databases without prior knowledge by converting natural language questions into SQL queries.In the existing works, several query processing frameworks are developed for retrieving data from the large database using advanced machine learning and deep learning techniques.But, the problems behind those works are computational burden, increased time consumption, high loss, lack of reliability, and accuracy.Therefore, the proposed work motivates to develop a simple as well as advanced feature extraction and machine learning models for an effective query processing.Here, the dictionary database (i.e.Text & Audio based SQL query formation) is created for system design and implementation.The proposed framework can handle both text and audio input data by extracting the features with the use of Lexical Text Data Analyzer (LTDA) and Mel-Frequency Cepstral Coefficients (MFCC) techniques respectively.Then, the extracted features are trained with the use of Bagged Bayesian Decision Tree (B2DT) classifier for an accurate query recognition.Finally, the performance of the proposed LTDA-MFCC integrated with B2DT model is validated and tested using several parameters.As we show, our model is able to improve an accuracy to 93.9% on the SQL query questions.In more detail, the accuracy, precision, recall, and F-score of instruction parsing reach 93.9%, 93.5%, 93.2%, and 93.4% respectively.Especially, it provides more convenient interactive means for accessing the databases through natural language statements in English language.