What does the survey comprise?
Statistics on patent applicants and applications from the Netherlands describe which enterprises are responsible for patent applications from the Netherlands. Data are broken down by a number of characteristics, for example economic activity, size class and technology area.
The target population includes all direct patent applications from the Netherlands to the European Patent Office (EPO) and to NL Patent Office, a division of NL Agency (NLOC). Patent applications to the EPO also include so-called EURO-PCT-filings. These are applications through the PCT to the World Intellectual Property Organisation (WIPO) designating to the EPO. Almost every patent application is linked with an enterprise from the General Business Register of Statistics Netherlands.
Enterprise. These are incorporated businesses or individuals. In official Statistics Netherlands terminology the enterprise is called enterprise group. An enterprise is an association of enterprises bound together by legal and/or financial links.
Date/year survey started
The study started in 2010 with the publication of results for the period 2000 to 2006.
Data for 2000 to 2006 were published in autumn 2010. The data for 2000 to 2005 are definite, those for 2006 are provisional. In 2011 (year t+5) data for 2006 will become definite. For future years, the publication strategy is as follows: in mid-year t+4 the dataset will be supplemented with data on year t. These are provisional figures. In mid-year t+5 the data for year t will become definite.
How is the survey conducted?
The study is based on registrations.
Electronic retrieval from registrations of patent applications.
At request of Statistics Netherlands, NLOC provided two registrations with all direct patent applications to the EPO and to the NLOC. Patent applications to the EPO also include so-called EURO-PCT-filings.
Not applicable (full retrieval from registrations).
Checking and correction methods
Based on the name and address of the applicant in the registration with patent information, a link was established with a statistical unit from the General Business Register. The statistical unit is the enterprise group. As the ownership of a patent generally lies at the highest level within the company, this is the most appropriate. In addition it is conceptually clearer and practically easier than establishing a link between patent applications and one of the other statistical units. It should be noted that universities and institutions are also covered by an enterprise group. Not all patent applications can be linked, e.g. applications by private persons, as private persons are not registered in the General Business Register.
The results for applicants (enterprises applying for a patent) are not weighted. If an enterprise filed patent applications in different technology areas or filed a patent application that can be classified into two (or more) technology areas, the enterprise counts fully for each technology area. The results for patent applications are weighted. If two or more enterprises are involved in a single patent application, the application counts for a fraction (depending on the number of applicants involved) in the breakdown by economic activity and size class. With regard to the breakdown by technology areas, if an application can be classified as relating to two (or more) technology areas, each technology area is allocated a fraction of the application.
The classification of patent applications by technology area uses a classification based on the International Patent Classification code (IPC code). The IPC codes are attributed by a patent office to patent applications from a technological point of view. An application is assigned an IPC code on the basis of its function or intrinsic nature of its field of application. The IPC is divided into sections, classes, subclasses and groups (in total the IPC has 70,000 categories). Statistics Netherlands compiled aggregates based on these categories.
Quality of the results?
The statistics are based on administrative sources that are enriched with information from Statistics Netherlands’ General Business Register. This process is not entirely flawless; in some isolated cases an application may be linked to the wrong enterprise or a patent application may erroneously not be linked.
First the link between patent applications and the General Business Register is established automatically using a number of algorithms that compare name and address in the two sources. Each automatically established link is assigned a score that can be interpreted as a report mark for the established link. The report mark can be considered to indicate the degree of correctness of the link. The score for the automatically established links is based on the ‘Largest Common Subsequence score’ (LCS score) and the ‘trigram score’. Both are outcomes of methods to compare names in string variables which are considered particularly suitable. The ‘trigram’ method mainly measures the similarity between string variables, and the LCS score measures the sequence equality. After completion of the process of establishing links automatically, a number of operations are executed to improve the quality of the links; not only checking and correcting links with low scores, but also linking unlinked patent applications manually.