As the Internet of Things (IoT) and similar technologies become more pervasive, it is important to consider how they might be used, not just to sell more stuff because your fridge says you’re out of milk and such-and-such a store now has milk on sale, but to properly assist people in the process of living. It seems an especially good idea to help people who are disabled, aging, or otherwise need assistance in their daily lives.

This book provides an overview of the authors’ research into using sensor networks and semantic web technologies to learn behavioral patterns, with an evident goal of providing support for people requiring assistance in daily activities.

The book consists of ten chapters:

(1) “Introduction” discusses the basic ideas of activity recognition, applications, and research trends.

(2) “Sensor-Based Activity Recognition” is on using sensors to monitor, model, and recognize “activities.” There are two main approaches: data driven, in which there is algorithmic modeling and recognition, and knowledge based, in which the models and rules are constructed by humans.

(3) “Ontology-Based Activity Recognition” covers how to use ontologies and ontological reasoning in support of activity recognition.

(4) “A Hybrid Approach to Activity Modeling” covers using data-driven model acquisition (that is, using sensor events to synthesize activity models) to produce knowledge-based systems.

(5) “Time Window-Based Data Segmentation” discusses using temporally labeled sensor events to identify activities and construct suitable ontologies for recognizing them.

(6) “Semantic-Based Sensor Data Segmentation” discusses a framework for combining ontologically represented knowledge with time-based activity recognition.

(7) “Composite Activity Recognition” explains that most activities are not single events (especially as signaled by a simple sensor) but are sequences of such events (consider a process such as “make tea” that integrates a number of simple activities as detected by sensor events into a single composite activity).

(8) “Semantic Smart Homes: Towards a Knowledge-Rich Smart Environment” gives an overview of the notion of a smart home.

(9) “Semantic Smart Homes: Situation-Aware Assisted Living” covers the technology that might support a smart home for assisted living.

(10) “Human Centered Cyber Physical Systems” covers some contrasting and evolving smart home prototypes.

Most of the chapters provide one or more algorithms in pseudocode, as well as the results of practical experiments.

Overall, the book is interesting and informative; however, among other difficulties, it feels padded. For instance, there’s a bar chart that takes up a third of a page to present exactly three values. In a couple of other places, there are tables where most of the data rows are identical except for the labels. One has 24 rows, and 20 of them are the same (again, except for the labels). At least one such table takes up an entire page. Information is repeated in many places. Many of the figures don’t seem to add much to the text, and more than a few are borderline illegible. On the other hand, in the bibliographic information, there seems to be a number of entries that are truncated (and they are in no clearly consistent bibliographic format).

Furthermore, one of the long tables has a puzzling summary row that contains the values 18/6, 16/8, 17/7, 14/1010, and 17/7 (the first number seems to be the count of entries L, and the second those of entries U).

It’s hard to determine an appropriate audience for this book. It is not really suitable for undergraduates. Graduate students outside the fields primarily covered, say, pervasive computing and the semantic web, are also likely to find it unpalatable. Active or potential researchers in the area are likely to be better served with primary research sources.