• Unit 9: Natural Language Processing

    This unit will delve into the full gamut of concepts and methods of natural language processing (NLP) and how it affects modern AI applications. The topics we will cover include what makes NLP hard, the grammatical structures of NLP, and various models, such as long short-term memory networks (LSTM), that have proved valuable in applications. Generative AI, such as ChatGPT, has become wildly popular.

    Completing this unit should take you approximately 6 hours.

    • Upon successful completion of this unit, you will be able to:

      • discuss the growing significance of natural language processing (NLP) in today's business applications;
      • explain why NLP has proved to be so difficult and the differences between processing, understanding, and generating natural language
      • explain the NLP concepts that can support understanding of text, including topic extraction, etc.;
      • explain the NLP concepts that support sentiment analysis;
      • detail the principles of LSTM and the NLP problems it can solve;
      • detail the principles of statistical learning methods to solve NLP problems;
      • describe how the most recent generative AI models work to create new content; and
      • describe the basics of ChatGPT and other applications that use large language models (LLMs).

    • 9.1: Foundations of NLP

      Natural language processing (NLP) has become more important in the rapidly changing internet. Human beings generate a lot of NL data, which has become interesting to companies selling products and those interested in market trends, social media monitoring, etc. We will look at the complexity inherent in NLP, including its ambiguity, the general knowledge of the world that gives NL its meaning, and the innate complexity of NL relative to other languages.

    • 9.2: How NLP Is Used

      This unit provides an overview of NL grammar and other concepts that aid NLP. There are many differences between NL understanding/processing and NL generation. Human beings generate a lot of NL data, and processing it to get actionable insights from that data is vastly useful to many decision-makers. Other applications include sentiment analysis, which decides if reviews are positive or negative based on content rather than a rating system.

    • 9.3: NLP Models and Methods

      Methods like LSTM combine advances in recurrent neural networks to solve problems like sentiment analysis of text. This kind of learning is also called deep learning. Other statistical learning methods can also be leveraged (short of a real understanding of the text) to analyze the influence of the NL corpus on human beings in different ways.

    • 9.4: Generative AI

      Generative AI models like ChatGPT have made a big splash in the marketplace. These advances are significant because interaction between humans and any intelligence mechanism is greatly facilitated. We will review the most recent advances in generative AI, including some applications and technology trends. We will discuss the ChatGPT platform through several use cases and applications that illustrate the platform's power. We will also study how large language models work to improve outcomes.