In short, I found this book of little or no value. Mr Graham opens with a brief description and history of O-O modeling and systems, and the benefits (and the pitfalls) they can bring. This section is relatively decent reference material, but is tainted with the author's unending litany of criticisms about earlier approaches and systems; he rarely seems to have words of praise for any system other than what he uses and promotes (SOMATiK, based on the SOMA Modeling Language). Next, the author talks in a fairly vague and pedantic manner about applying O-O concepts to improve the quality and speed of requirements engineering. Once you've read this, you expect a strong presentation of a breakthrough way of doing requirements engineering development. However, the remainder of the book is short and anti-climactic, and hardly corroborates his claims for the benefits of O-O in requirements engineering and development. Finally, Mr Graham includes some theoretical material, e.g. on requirements provability, in appendixes, perhaps to show off his mathematical prowess.
I'm not in a position to dispute the author's credentials, or his claims of having developed some excellent real-world systems. It's just that Mr Graham fails to make his case, and, along the way, does not teach us much about how to actually perform requirements engineering in the real world. I also have my doubts as to whether the SOMATiK system that he over-promotes has much currency (see http://www.techweb.com/encyclopedia/defineterm.jhtml?term=SOMATiK, which defines the term as "An earlier CASE tool from the former Bezant Ltd., Wallingford, Oxfordshire, England, that was known for its acceptance of specifications in text form ..."). Finally, if this were not enough, the author's criticisms of other people's work are both unprofessional and tedious.