Roy Zimmermann wrote:
>
> I'll suggest another field of data derived from US facepages that has > contained significant numbers of errors over the years, viz., Cited > Reference Patent Numbers. ... I consistently found between
> 10-20% of the cited reference US patents numbers contained errors, > either in the patent number itself, or in a mismatch between the cited > patent number and the original class or inventors names found after > the cited references. Clearly, I couldn't take US cited reference > patent numbers at face, or facepage, value!
>
> Does this indicate a quality problem in the production of US patents? > Duh!? But this is not soley the PTO's problem, nor the various
> printing contractors used over the years by the PTO to produce the
> printed US patent specs. Consider how rarely you see Certificates
> of Correction with US patents compared to how often you find obvious
> errors in both the facepage and claims of US patents. The
> applicants contribute to the proliferation of errors as does the PTO
> and its internal processes and subcontractors.
Regardless of the type or over all quality of the citing literature
(including patents), the authors of those documents are notoriously
error prone in writing citations. Citation searching of any kind must
be done by expanding all possible forms of the author's name and, for
citations of the non-patent literature, examination of various
permutations of the numbers involved. Of course, if citations to a
specific patent number are involved, good luck -- there's no way to
determine which patent numbers are correct or incorrect without
examining the original document (or a trusted surrogate).
At least for literature citations, a combination of the volume number
and date of publication provides some sort of linking correct and
erroneous citations. However, I've even seen literature citations where
names have been misspelled, including one notable instance where all 5
authors names were misspelled in a citation, and in a notable journal no
less.
-- Bob Buntrock
Buntrock Associates, Inc.
This archive was generated by hypermail 2b30 : Fri Feb 14 2003 - 11:56:37 GMT