Design and Implementation of Neural Processor for Parsing Manufacturing Query Language

• Published in: International journal on computer science and engineering Vol. 7; no. 12; p. 142
• Main Authors: Naik, Girish R, Raikar, V A, Naik, Poornima G
• Format: Journal Article
• Language: English
• Published: 01.12.2015
• Subjects: Computer programs; Format; Matlab; Microprocessors; Neural networks; Query languages; Software; Trees
• ISSN: 0975-3397

Practically, all the approaches employed for parsing with natural languages use some or other type of neural network architecture and some typical statistical function for obtaining a parsing decision. In parsing with neural networks an incremental tree is usually obtained by using a set of rules for connecting a possible parse tree to the previously obtained incremental tree. In the current work, linguistic data is mapped to corresponding part-of-speech tags, which are then converted into a set of binary input vectors for each sentence. The tags have relationships with their neighbours which are modeled by the neural processor. When input is given to the neural processor, these relationships are analyzed and the string with the correct placement of parts-of-speech tag is output as syntactically correct else is declared as syntactically incorrect. A single layer network with back propagation is employed which utilizes a method based on minimization of error between the desired and actual activation of output nodes. A model is developed for dynamically accepting a query in natural language in the presentation tier of multi layered architecture which is processed and sent to the middle tier interfaced with R Software and MatLab for training the neural network and testing the query input by the user. The requisite Excel file in CSV format are generated and processed in the data layer. The entire approach is rendered generic and can be applied to similar cases containing the training data in the requisite format in Excel file. The confusion matrices generated by both the softwares are compared for judging the accuracy of classification.