Do patent citations indicate knowledge linkage? The evidence from text similarities between patents and their citations

•We validate the basic assumption of the patent citation analysis method: patent citations can indicate knowledge linkage.•Patents are generally more relevant to their citations (cited patents) than other patents in text similarities.•Patent applicant citations and examiner citations can indicate knowledge relevance and knowledge diffusion.•Examiner citations indicate knowledge linkage better than applicant citations, especially for the part of patent claims.•Preferably, examiner citations are a good indicator of knowledge linkage rather than an incomplete and noisy indicator.

Whether patent citations indicate knowledge linkage is still a controversial issue, which is very important for the widespread use of the patent citation analysis method. We hypothesize that there exists technological knowledge linkage between patents and their citations, and that the linkage can be detected through measuring text similarities between them. To test the hypothesis, we selected citing-cited patent pairs as the observation group and selected patent pairs without citing-cited relationship as the control group. Using the VSM with WF-IDF weighting method, we calculated text similarity values of the two groups. Through comparing text similarity values between the two groups, we validate that in the vast majority of cases text similarity values of citing-cited pairs are much higher than those of non-citing-cited pairs. The study in nano-technology field shows that the above results are the same, although patents in the same technological area are more relevant than in different technological areas. Furthermore, by comparing text similarities between applicant and examiner citing-cited pairs, the results show that in more cases examiner citations indicate knowledge linkage a bit better than applicant citations. Preferably, examiner citations can be regarded as not only the supplement of applicant citations but also the more important technological background and the prior art closely related to the patents. Compared to applicant citations, examiner citations are a good indicator of knowledge linkage rather than an incomplete and noisy indicator. In short, the results suggest that most certainly patent citations can indicate knowledge linkage, and more likely examiner citations can indicate knowledge linkage a bit better than applicant citations, especially for the component of patent claims. Therefore, we accept the hypothesis that patent citations can indicate knowledge linkage, which is the basic assumption of the patent citation analysis method.