Machine Learning and AI with Simple Python and Matlab Scripts

6ANNs for Natural Language Processing

6.1 Introduction

As one of the fastest growing applications of AI, natural language processing (NLP) aims to allow computers to understand, interpret, and manipulate human language. In this chapter we present an NLP application that attempts to identify a possible and typically previously unrevealed relationship between words presented in a content (e.g. speech, written document or text message) and external events happening in various seemingly unrelated venues. Examples of this type of application include a speech given by a politician followed by rising oil prices, or a famous singer praising a specific brand of shoes in her concert causing an increased demand on foreign labour in the shoe industry. NLP is often seen as an intersection between linguistics, computational disciplines and engineering and involves algorithmic analysis of natural languages [63].

Inputs to NLP algorithms are typically a set of features that are generated based on information collected from external sources or various existing platforms. Information to be explored by an NLP application could be verbal or written content, from which the input features could be derived using various approaches (e.g. a set of key words in specific content such as text messages from a prominent person in a given field). For example, a given set of interaction fragments (e.g. a chain of emails, voice messages or text message exchanges) can be analysed by an NLP algorithm to explore ...