Understanding spoken and written communication can be difficult, even for humans. Creating linguistic processes and algorithms for computers is extremely challenging. Natural language processing (NLP) technology is more common than most people realize: it’s widely used in web searching, commerce and finance, social media, and digital communication. Furthermore, there has been a burst of publications, research, and development related to NLP in recent years, from commercial products like Amazon’s Alexa and IBM’s Watson, enabled by artificial intelligence (AI), machine learning (ML), and program development tools, and
libraries like Python’s Natural Language Toolkit (NLTK) and the Keras deep learning library.
Among the myriad publications, blogs, and web tutorials about Python for NLP, Mathangi Sri’s book stands out for its focus on real-world use cases. The author is well known in India; she is head of Data Science at PhonePe, a developer of widely used mobile payment and service applications. Her approach to the subject covers four major NLP domains: customer service, online reviews, banking and finance, and virtual assistants. In each of the chapters covering these areas, Sri explains the unique data types and methods used for extracting meaning and context from text and voice data, using a variety of Python libraries.
Sri assumes fluency with Python, providing code fragments and function details; full source code is available from the Apress GitHub website. Readers with
entry-level Python skills should spend some time with introductory programming resources before trying to understand and exploit the book’s code examples.
Each of the book’s four chapters describes multiple approaches to the area of analysis, from simple or “classic” methods to more complex
ML-based solutions. For example, the chapter about online reviews text starts with a basic lexicon methodology, simply identifying and counting sentiment words and phrases. A second suggested approach uses a rules-based method to improve program performance,
and a more advanced neural net model is later proposed and detailed.
Apress offers a printed version of the book, but the text includes many helpful embedded uniform resource locator (URL) web references, so readers might prefer
the available digital edition that allows interactive access to
those links. Many books on this topic provide elementary guidance
and simplistic source code, instructive for beginners but perhaps
not appropriate for experienced developers’ practical goals.
Sri’s contribution fills that instructional gap with relevant
and usable Python code examples.
More reviews about this item: Amazon