Tuesday, September 12, 2006

Data on Inventors

Patent data (and data on inventors) has often been used in economic research. However, it is difficult to track inventors over time due to name changes, different spellings, and non-unique names. Since over 3.1 million patents have been issued by the USPTO between 1963 and 1999 (the patents included in the NBER patent data file), this is an impossible hurdle to overcome manually.

Manuel Trajtenberg, Gil Shiff, and Ran Melamed have recently produced a paper that tackles the "who is who" problem in this dataset. Thus, they have created a even richer patent dataset which allows for the tracking of inventors (including the inventor's residential location as well as that of the firm and relevent patent information, such as citations, etc.) The paper presents the methodology used to overcome this problem; it looks very much like it could be used in other datasources with similar problems.

I don't see that they have yet made their data available for use, though the paper seems to indicate that it should be at some point. This is a great project and the authors should be commended for creating what should become a very valuable data set. Hopefully, it will be out there soon enough, and we'll see a bunch of interesting new work coming out of it....


Post a Comment

Links to this post:

Create a Link

<< Home