“By combining statistical analysis, natural language processing, artificial intelligence, and business judgement, Peter demonstrates how sentiment and reputation can be measured, valued, predicted, and ultimately managed with greater confidence and accuracy.”
- Alberto López Valenzuela, Former Founder & CEO of alva, author of The Connecting Leader (Lioncrest, 2018)
“Particularly noteworthy is its integration of established statistical methods with the latest advances in large language models, providing a timely perspective on the future of reputation analytics.”
- Carolyn Phelan, Associate Professor, UCL
“Now the world is ready for this, and I recommend you read this important text.”
- Richard G. Fleming, MBA MBCS CITP – veteran CTO, InnovateUK UKRI - IT Director
Can reputation be measured directly? Yes! The novel idea in this book is that “reputation” is a time series of numerical “sentiment” values. Each one summarises, on a daily basis, what people think about a target organisation, brand or product. The ideas here present reputation as a tangible asset that can be measured, quantified, and used to manage risk and steer company policy.  They are not just a collection of mathematical and statistical techniques: they represent a thought-provoking mindset change in business intelligence, and a paradigm shift in how reputation is calculated. Asking only a handful of people what they think about the target on one day only is replaced by asking hundreds or thousands of people, every day.
The statistical and mathematical properties of sentiment and reputation are explored in depth using the R statistical language at an undergraduate level. The progression from texts sourced from the internet to reputation is encapsulated in the STAR – Statistics, Text, Analytics, Reputation – pipeline. STAR works, for each target, by extensive text mining, calculating a numerical score for each, and combining them into daily numerical sentiments. Sequential sentiment values – the target’s profile – is the world’s view of the target, and reflects its reputation. Large Language Models are used for both measurement and, with some words of caution, reputation risk management.
Readers will be able to:

• use and adapt the R code in each chapter;
• learn how to use traditional and Large Language Models for sentiment analysis and corporate decision making;
• monitor and forecast reputation;
• apply reputation risk controls.