In the complicated world of patent searching a new tool is available to help analyse data in patent search reports.
When patent specifications are published many countries publish search reports listing earlier documents or “prior art”, mostly patents, that are of some similarity. This information is vital to the applicant, of course, as it will determine if the invention will be allowed.
Others will find the search reports of interest. Many countries publish this data on the first, A document rather than the later granted B document. This means that if an application is published there is a ready source of information on possibly damaging prior art to help those wanting to stop the patent or simply interested in new technology.
The normal approach is to look at each patent authority’s document for the same invention in turn. Now the Trilateral Offices (the USA, Japan and the European Patent Office) have put up a free tool, the Common Citation Document site.
You can enter a patent number and see different reports listed one after the other on the same screen. Look for example for US7479949, widely regarded as the main patent for the iPhone®.
It occurs to me that it would be useful to have a method to highlight, perhaps as an optional page, those citations that turn up the most times – so that if a patent is cited by say three patent offices it suggests that they are likely to be of special relevance.
What I found really fascinating at the site is the “Inspector”, where on the right hand side of the page the classifications allocated by the different offices appear one after the other. Classifying patent documents must be very difficult, and sometimes offices allocate different classes to those by other offices.
Take for example EP2015321, which is for magnetic cores. The European Patent Office, China and Japan all gave the same 5 International Patent Classification (IPC) classes.
The USA, on the other hand, gave 4 classes, only one of which was included within the 5. This may be related to the fact that at present the US office allocates the IPC by using a computerised concordance from their own classification scheme.
In addition there is EC (European Classification) data. This is independently applied by staff at the European Patent Classification to the invention, often with special, detailed subsets of the IPC. There are 8 EC classes for the invention – only 2 of which fall within the same classes as found by the offices (both being in the EPO/China/Japan listings).
Or take EP2097145, for a toy vehicle. The US gives a different, more specific class from the other offices’ class, while Canada adds a second class to the others. EC gives 3 classes – all different from those 3 different classes. You could, admittedly, use the broad A63H17/00 and ask for everything within it which in this case would work, but it would pick up a lot, and classification is meant to narrow down what is found.
These two examples were chosen entirely at random, and give food for thought. Maybe a large sample should be analysed ? Each search, at least in unfamiliar subject areas, should include carefully checking how the classes have been used by the offices.
These results may not be typical, and I do appreciate the great problems in classifying in a pressurised work environment. Nevertheless they do show that patent searching is a complex art and not, as some think, a simple, automated procedure.