Patents, Innovations and Citations

Citation

Oppenheim, C. (2003), "Patents, Innovations and Citations", Journal of Documentation, Vol. 59 No. 3, pp. 369-371. https://doi.org/10.1108/00220410310472563

This is a book summarising many years’ research by the authors, together with various collaborators, over the period 1990‐2001, together with introductory text and concluding chapters. With one or two exceptions, the research is focused on the use of patent citation data for economic analysis of countries, of so‐called spillover of research from one organisation to another, and of the impact of university research.

The use of patent citation data is a complicated and controversial area. Those readers who know of the controversies surrounding the use of journal citation data will be aware of many of the issues that arise. The situation is even more complex in the case of patent citation data because citations in patent specifications come from two sources (the patent examiner and the applicant) and not one (the author) as in the case of journal articles. Furthermore, the motivations for citing in a patent document are very different for both the applicant and the patent examiner, and both their motivations are different again from the motivations that authors of journal articles have for citing references. In short, patent citation analysis is a dangerous minefield.

Sadly, the authors and their collaborators are totally unaware of the dangers of uncritical use of patent citation data. The rare times they even consider the question in this book, they reassure the reader that citations made in patents are reliable. They provide no supporting evidence. The reason why they supply no supporting evidence – or indeed consider the possibility that this is a minefield – is that they are, astonishingly, clearly totally ignorant of the considerable literature that exists on the pros and cons of patent citation data for bibliometric or competitive intelligence purpose.

A simple search on Web of Science or in LISA would throw up dozens of articles critically evaluating the use of patent citation data. The authors of this book note not one. In short, they have pursued a blinkered approach to their topic. One searches in vain for the names of Stu Kaback, Edlyn Simmons, myself and the many other authors who have explored the use of patent citation data in depth. Indeed the authors make the outrageous claim that the first use of patent citation data was in 1983 – about ten years after the first use I am aware of, and many more years since Garfield first suggested that it could have use.

And what a structure they have built on such shaky foundations. The book is replete with complex mathematical formulae and correlation coefficients. Much of the mathematics was far too complex for me to comprehend, but I did get hold of one statement that was repeated several times in different chapters, that for the purposes of their calculations, they have assumed that log(1+a/b) is approximately equal to a/b. Try as I can, I cannot find any value of a/b for which that even holds vaguely true. If this is an example of the authors’ mathematical knowledge, I have no confidence in the more complex material, which they also present.

Because the various chapters, which are closely based on, or are identical to originally journal articles, have not been properly edited, there is significant repetition of text between them, and some use dated data.

The authors also misunderstand the purpose of patents (which is to prevent others from making using or selling the invention, and not, as claimed so that the patent owner can make commercial use of the invention), and do not seem to be aware of the difference between early‐published applications and final granted patents. The claim that US patents are always of a high standard will raise a hollow laugh from all those experienced in patent searching, and the authors seem totally unaware that there are patent systems outside the USA.

Only now and again do the authors report research where they followed up their results by talking to the researchers involved – and never seem to have talked to patent examiners about the motivations for citing. They also seem to be quite unaware of the many commercial databases that they could have used rather than laboriously building their own database of patent citation records. They also seem unaware of the work by Line and his co‐workers on the question of diachronous versus synchronous citation data, a topic they appear to have discovered for themselves in the last year or so.

To summarise my view of this book is easy; I quote from a recent review article I wrote on the use of patent citation data:

It is remarkable that so much work has been undertaken based upon unproved or shaky foundations. The most extreme example is … analyses of patent citation data using complex algorithms and correlation coefficients … based on their claim, without evidence, that patent citations are more robust and less noisy than citations in the scientific literature. They also claim that the non‐patent citations made by the examiner do not carry the legal weight of patent citations, a total misunderstanding of patent law. Finally, they base some of the analysis on patent classification symbols assigned by patent examiners, without realising the unreliability of these assigned symbols.

The book cannot be recommended as it is based on an uncritical and flawed understanding of the role of patent citations and, as far as I can tell, on shaky mathematics. My recommendation to the authors is simple: read the library and information science literature.