Deep Learning Neural Network Model for Natural Language Processing and Application of Sentiment Analysis to Define the Text Content : A Case Study: BBC Christmas Cooking Recipe
Recent advances in machine learning have allowed traditional software application domain to integrate AI capabilities. Machine learning, especially deep learning, differs from traditional software engineering in which its implementation is heavily dependent on data from the external world. Yet, there’s still a lack of practices for implementing the deep learning model in software development due to the need for curating and processing voluminous, heterogeneous, and often complex data. Since deep learning is able to compose a new model from sub-models and reuse pre-trained models, this “transfer learning” technique can enhance the model if it is used in a strategic way. This research result shows that eliminating unnecessary multiplication by modifying input layers and reducing noise in the vocabulary while training the multilayer perceptions are those strategies with efficient applications of technical debts. In order to implement the deep learning model in software engineering domain, curating voluminous data with proper “transfer learning” technique is a key to increase the accuracy of the model. This paper experiments how to efficiently build and train the deep neural network model with complicated natural language data to enhance the accuracy of the model. It introduces the basic mathematical background of how the model works, and shows two different ways of improving the accuracy for the deep learning model. As a case study, the method descriptions of Christmas cooking recipes is reviewed.
iv, 25 p.
U.S. copyright laws protect this material. Commercial use or distribution of this material is not permitted without prior written