the overlap of u.s. and canadian patent literature with journal literature literature with journal...

4
Judith Allen and Charles Oppenheim*, Centre for Information Science, The City University, London The Overlap of U.S. and Canadian Patent Literature with Journal Literature Summary The paper describes a study of the overlap of Canadian and U.S. patent literature and journal literature. A random sample of 300 U.S. and 100 Canadian patents published in 1968 were examined. 6.0% of the U.S. patents and 11.0% of the Canadian patents also appeared in the journal literature. 6.1% of the chemical patents, 6.0% of the mechanical patents and 11.3 % of the electrical patents also appeared in the journal literature. The results confirm the need to search both the journal and the patent literature for com- prehensive literature coverage. L’expost dtcrit une etude sur le probl?me du recouvrement de la litterature brevets au Canada et aux Stats-Unis. Un echantillon de 300 brevets US et 100 brevets canadiens a ete examine. 6% des brevets US et 11% des brevets canadiens ont paru egalement dans la litterature des periodiques. En outre, 6,1% des brevets de chimie, 6% des brevets de mecanique et 11,3 % des brevets d’electricite ont paru dans des periodiques. Ces resultats confirment la necessite de proceder a des recherches non seulement dans la litterature de periodiques, mais egalement dans la litterature brevets, pour arriver a une couverture totale de la litterature. Der Beitrag beschreibt eine Studie iiber das Problem der Uberschneidung der kanadi- schen und US-Patentliteratur mit der Zeitschriftenliteratur. Es wurde eine Stichprobe von 300 US-Patenten und 100 kanadischen Patenten untersucht. Davon erschienen 6% der US-Patente und 11% der kanadischen Patente such in der Zeitschriftenliteratur. Ferner erschienen 6,1% der Chemiepatente, 6,0% der Patente Mechanik und 11,3 % der Patente Elektrotechnik in Zeitschriften. Diese Ergebnisse bestatigen die Notwendigkeit, nicht nur in der Zeitschriftenliteratur, sondern such in der Patentliteratur zu recherchie- ren, urn eine umfassende Literaturabdeckung zu erreichen. Introduction Although it is known that patents are an important source of scientific and technical information, the useage of patent literature by most scientists remains low. There are many possible reasons for this lack of use, but the principal ones appear to be as follows: 1. Patents are written in obscure terminology; 2. Ignorance about patent law and how the patent literature can be obtained; 3. Inadequacy of many of the standard secondary services in their coverage of the patent literature means that few patents are drawn to the attention of a searcher; 4. The numbers of libraries holding patent collections are low; 5. It is thought that the information contained in patents will be brought to the attention of readers in journal articles if the information is at all im- portant. Although many of these reasons are justified, the last one is certainly not. Work emanating from this Centre (1) has demonstrated how important techno- logical developments appear in the patent literature alone, or only appear in journals many years later. Liebesny has cited other examples (2). All the exam- pies quoted in these publications refer to specific in- ventions, and it could be argued that although in cer- tain cases the patents are not duplicated by journal articles, in general the overlap between patent and journal literature is reasonable. Work carried out in this Centre (3) has shown that 40-60% of technical reports in the field of chemistry are duplicated by journal articles and an equivalent figure might be expected for patents. In 1974 Liebesny and his co-workers studied the overlap of a random sample of 1058 British patents with the journal literature (4). The patent specifica- tions were published in 1962, 1965 and 1968. Name (author) searches were carried out in the names of the inventors through Chemical Abstracts, Engineering Index and Electrical and Electronic Abstracts. The main results obtained by Liebesny’s team are given in Table 1, and they demonstrate an overlap rate of 3-9% depending on subject-matter. The analysis of time differences between date of publication in a journal article and date of publication of the patent or date of application for a patent showed that a peak of journal publishing occurred one year after application for the corresponding patent. * To whom correspondence should be addressed. wodd Patent Information 1 (1979) No. 2 Oppenheim - U.S. and Canadian Patent Literature 77

Upload: judith-allen

Post on 02-Jul-2016

217 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The overlap of U.S. and Canadian patent literature with journal literature literature with journal literature

Judith Allen and Charles Oppenheim*, Centre for Information Science, The City University, London

The Overlap of U.S. and Canadian Patent Literature with Journal Literature

Summary The paper describes a study of the overlap of Canadian and U.S. patent literature and journal literature. A random sample of 300 U.S. and 100 Canadian patents published in 1968 were examined. 6.0% of the U.S. patents and 11.0% of the Canadian patents also appeared in the journal literature. 6.1% of the chemical patents, 6.0% of the mechanical patents and 11.3 % of the electrical patents also appeared in the journal literature. The results confirm the need to search both the journal and the patent literature for com- prehensive literature coverage.

L’expost dtcrit une etude sur le probl?me du recouvrement de la litterature brevets au Canada et aux Stats-Unis. Un echantillon de 300 brevets US et 100 brevets canadiens a ete examine. 6% des brevets US et 11% des brevets canadiens ont paru egalement dans la litterature des periodiques. En outre, 6,1% des brevets de chimie, 6% des brevets de mecanique et 11,3 % des brevets d’electricite ont paru dans des periodiques. Ces resultats confirment la necessite de proceder a des recherches non seulement dans la litterature de periodiques, mais egalement dans la litterature brevets, pour arriver a une couverture totale de la litterature.

Der Beitrag beschreibt eine Studie iiber das Problem der Uberschneidung der kanadi- schen und US-Patentliteratur mit der Zeitschriftenliteratur. Es wurde eine Stichprobe von 300 US-Patenten und 100 kanadischen Patenten untersucht. Davon erschienen 6% der US-Patente und 11% der kanadischen Patente such in der Zeitschriftenliteratur. Ferner erschienen 6,1% der Chemiepatente, 6,0% der Patente Mechanik und 11,3 % der Patente Elektrotechnik in Zeitschriften. Diese Ergebnisse bestatigen die Notwendigkeit, nicht nur in der Zeitschriftenliteratur, sondern such in der Patentliteratur zu recherchie- ren, urn eine umfassende Literaturabdeckung zu erreichen.

Introduction

Although it is known that patents are an important source of scientific and technical information, the useage of patent literature by most scientists remains low. There are many possible reasons for this lack of use, but the principal ones appear to be as follows: 1. Patents are written in obscure terminology; 2. Ignorance about patent law and how the patent

literature can be obtained; 3. Inadequacy of many of the standard secondary

services in their coverage of the patent literature means that few patents are drawn to the attention of a searcher;

4. The numbers of libraries holding patent collections are low;

5. It is thought that the information contained in patents will be brought to the attention of readers in journal articles if the information is at all im- portant. Although many of these reasons are justified, the

last one is certainly not. Work emanating from this Centre (1) has demonstrated how important techno- logical developments appear in the patent literature alone, or only appear in journals many years later. Liebesny has cited other examples (2). All the exam-

pies quoted in these publications refer to specific in- ventions, and it could be argued that although in cer- tain cases the patents are not duplicated by journal articles, in general the overlap between patent and journal literature is reasonable. Work carried out in this Centre (3) has shown that 40-60% of technical reports in the field of chemistry are duplicated by journal articles and an equivalent figure might be expected for patents.

In 1974 Liebesny and his co-workers studied the overlap of a random sample of 1058 British patents with the journal literature (4). The patent specifica- tions were published in 1962, 1965 and 1968. Name (author) searches were carried out in the names of the inventors through Chemical Abstracts, Engineering

Index and Electrical and Electronic Abstracts. The main results obtained by Liebesny’s team are given in Table 1, and they demonstrate an overlap rate of 3-9% depending on subject-matter.

The analysis of time differences between date of publication in a journal article and date of publication of the patent or date of application for a patent showed that a peak of journal publishing occurred one year after application for the corresponding patent.

* To whom correspondence should be addressed.

wodd Patent Information 1 (1979) No. 2 Oppenheim - U.S. and Canadian Patent Literature 77

Page 2: The overlap of U.S. and Canadian patent literature with journal literature literature with journal literature

Liebesny also made an analysis of the number of mul- tiple publications from his sample of patents, and found that in six cases, two journal articles appeared, in three cases three journal articles appeared and in one case four journal articles appeared.

Table 1: British specifications retrieved in other forms of literature

1962 1965 1968 Total

Chemical Investigated 91 95 127 313 Retrieved 9 10 8 27 % retrieved 9.9 10.5 6.3 8.6

Electrical Investigated 49 89 74 212 Retrieved 5 4 7 16 % retrieved 10.2 4.5 9.5 7.5

Mechanical Investigated 143 181 209 533 Retrieved 3 12 3 18 % retrieved 2.1 6.6 1.4 3.3

% of total retrieved 6.0 7.1 4.4 5.8

Liebesny’s work only covered British patents, which may conceivably be untypical. We therefore decided to extend his study to U.S. and Canadian patents in order to see if the results obtained by Liebesny’s team were general.

Methodology

A random sample of 300 US and 100 Canadian patents published in 1968 was taken. Both US and Canadian patent specifications are divided into three broad groups when published: General & Mechanical (GM), Chemical (C) and Electrical (E). Although US patents are issued weekly and each week’s output is arranged in numeric order within these three groups, Clark (5) has shown that one can still select a random sample of US patents by taking every 500th specification. We in fact selected every 50th specification, i.e. those patents whose numbers ended in 00 or 50. Table 2 shows the subject breakdown of the patents investigated.

Table 2: Subject breakdown of patents studied

U.S.A. CANADA Total for Total for

Sample 1967 Sample 1968

Mechanical 59% 57% 46% 62% Chemical 24% 25% 34% 26% Electrical 17% 18% 20% 12%

The total US proportions were ascertained from pub- lished figures (11). The total Canadian proportions were taken from official figures (6). Table 2 demon- strates that the sample taken follows the same general trends of proportions as the whole year, though not quite as closely as would have been desired.

Details of the inventors and titles of the patents selected were taken from the Official Gazettes of the U.S. and Canadian Patent Offices. An author search was then carried out on all the inventors named through the three services used by Liebesny and his co-workers. These were searched for the period 1963- 1977; for part of this time period, these services are available on-line. On-line searching was carried out using COMPENDEX and Chemical Abstracts Conden- sates on ESA Recon for some of the searches.

When hits were found, we calculated the time lag between priority date of the patent application or publication date of the patent with the publication date of the journal article. We also calculated what proportion of the patents in our sample were them- selves noted by the abstracting journals. We also checked on multiple articles emanating from one patent.

Results

Table 3 summarises our results and Table 4 compares our results to those of Liebesny and his co-workers. It should be borne in mind in such a comparison that Liebesny’s definition of C., MG and E in UK patents may not be quite the same as the U.S. or Canadian Patent Office’s definitions.

Table3: Specifications retrieved as other forms of literature

Chemical Investigated Retrieved % retrieved

Mechanical Investigated Retrieved % retrieved

Electrical Investigated Retrieved % retrieved

% of total retrieved

USA Canada

179 34 7 6 3.9 17.6

70 46 5 2 7.1 4.3

51 20 5 3 9.8 15.0

6.0 11.0

Total

213 13 6.1

116 7 6.0

71

Ii.3

7.3

78 World Patent Information 1 (1979) No. 2 Oppenheim - U.S. and Canadian Patent Literature

Page 3: The overlap of U.S. and Canadian patent literature with journal literature literature with journal literature

Table 4: Comparison of our results with Liebesny’s

% chemical patents .retrieved

% mechanical patents retrieved

% electrical patents retrieved

% total patents retrieved

U.K. U.S.A. Canada

8.6 3.9 17.6

3.3 7.1 4.3

7.5 9.8 15.0

5.8 6.0 11.0

Size of sample 1058 300 100

These results are similar to those obtained by Lie- besny et al. and demonstrate the generality of the non- overlap of journal and patent literature. Canadian patents seem to be duplicated more in the journal literature than British or U.S. patents, but the Cana- dian sample size was smaller.

We found the average number of journal articles covering the same subject matter as a patent to be highly consistent at 1.4 per patent, irrespective of sub- ject matter. Liebesny found a very similar figure for his patent sample of 1.3 articles per patent. It must be borne in mind, of course, that other articles, reports, etc. might appear in publications not abstracted by the secondary services we used, so these figures are minimum figures.

The vast majority of the duplicated items appeared in the journal literature; we noted four conference papers and no reports amongst the items. Mtrltiple publication items were journal articles only; we found seven cases of two journal articles emanating from a patent and one case of three journal articles emanating from a patent.

Table5 Time-lag of publication in other forms of literature in years from patent priority date

Year from Priority 0 1 2 3 4 5 6 7 8 9 10 II+

U.S.A. 3512432010 1 1 Canada 0140211301 0 1

TOTAL 3652643311 12

Table6: Time-lag of publication in other forms of liter- ature in years from patent publication date

Year from publication -6 -5 -4 -3 -2 -1 0 1 2 3 4 5+ date U.S.A. 121454211110 Canada 011332111001

TOTAL 132786322111

Tables 5 and 6 summarise our results for time-lag data between the priority date for the patent and

their subsequent publication in another form of litera- ture and also for the time lag with date of publication of the patent.

These results demonstrate that most of the items in other forms of literature appear l-4 years after the priority date of the patent, a result identical with that found by Liebesny and his co-workers. It was no sur- prise to find that the items in other forms of literature came out similarly in relation to publication of U.S. and Canadian patents as Liebesny et al found for Brit- ish patents. Most. of the items in our sample appeared l-3 years before the patent was published, and 19% appeared after the patent appeared. Liebesny et al found that 38% of their sample appeared l-3 years be- fore the patent was published, and 16% appeared after the patent was published. Of course Canada and USA are “slow publishing” patent issuing countries, so it is not surprising that patents often appear after an equi- valent journal article.

We also calculated the time delays between priority date and publication of the granted patent. For USA, most patents appeared three or four years after priority date and for Canada most appeared four or five years after priority date.

Since this research was carried out, a report has appeared describing a similar study carried out on 220 1967 U.S. patents and 215 1972 U.S. patents (11). Both author and subject-matter searches were carried out using the same abstracting services as we employed. Of the 435 patents searched, 44 (10.5%) were the sub- ject of publication in the non-patent literature. The overlap rates were Chemical 24.3%, Electrical 5.0% and Mechanical 5 .O%. The overall results are somewhat higher than our figures, with a markedly higher overlap rate for chemical patents and a lower rate for electrical and mechanical patents. We are at a loss to explain the differences between our results. A study of the time interval to disclosure confirmed our results that for patents disclosed in non-patent literature, publication occurred for the most part prior to the issuance of the U.S. patent. A study of multiple disclosure showed that 1.7 of the 44 patents resulted in one non-patent item, but that multiple disclosure occurred for the re- maining 27 patents. On average, each patent with non- patent counterparts resulted in 3.1 non-patent publica- tions, but this figure was distorted by two chemical patents, one of which resulted in 12 non-patent refer- ences and one of which resulted in 20 non-patent refer- ences.

Turning to the abstracts journals, we found that Chemical Abstracts was fastest at noting equivalent journal articles. Over half the articles were abstracted within three months of their publication. Electrical and

World Patent Information 1 (1979) No. 2 Oppenheim - U.S. and Canadian Patent Literature 79

Page 4: The overlap of U.S. and Canadian patent literature with journal literature literature with journal literature

Electronic Abstracts came next; half the abstracts it noted appeared within four months of the items’ ap- pearance. In the case of Engineering Index, half the items abstracted appeared within ten months of the items’ appearance.

None of the original patents in the sample were noted by Engineen’ng Index or by Electrical and Elec- tronic Abstracts. This is clearly a serious deficiency of both these abstracting services. 60 of the 70 U.S. chemical patents (86%) were noted by Chemical Ab- stracts, but none of the 34 Canadian chemical patents were picked up by Chemical Abstracts.

The figure of 86% for U.S. chemical patents is somewhat lower than results obtained by Kaye (7) and by Oppenheim and his co-workers (8,9). The figure for coverage of Canadian patents is very disappointing.

Conclusions

The results show that it is possible to extend the results of Liebesny’s study on British patents to U.S. and Ca- nadian patents. Although Canadian patents show a higher value for the number of subsequent publica- tions, over the sample as a whole the results agree fairly clo,sely. That is: only a very small proportion of Brit- ish’, U.S. and Canadian patents (less than 10%) eventu- ally give rise to publications in other forms of litera- ture. This seems an important factor for any scientist to consider when he is attempting to find information on a particular subject.

Having said that there is only a minimal amount of publication subsequent to the publication of a patent, the next factor to consider is: where such publications do arise what are the time differences involved. We have shown that with the patents studied there is a peak in publication one year after priority date. This is to be expected as once a patent has been applied for, there should be no problems with getting it granted, so long as no publications were made prior to application. Thus any inventor who wants to publicise his invention is quite likely to do so as soon after the priority date as possible.

Thus, if a scientist on searching for information on a particular subject happens to cover the journals in which an inventor has published the details of his in- vention he will be fortunate in covering the area more extensively. This will mean that the need to examine the patent literature itself will then be removed. How- ever, it is more likely that he will not search all the relevant journals, but will more probably cover the major abstracting services in the field. Thus it is also of interest to examine the time lag between publication of the patents or their equivalent articles, and the publica-

tion of the abstracts of these in abstracting journals. The results show that only U.S. chemical patents were abstracted directly by Chemical Abstracts - no patents were found in the two other main abstracting services studied. The delay between publication of the articles and their abstracts in abstracting journals reflects the efficiency of the particular abstracting service. The results seem to show that Chemical Abstracts also per- forms best here with at least half the abstracts appear- ing within three months. Electrical and ElectronicAb- stracts is next best, and Engineering Index poorest, with no abstracts appearing until five months after publication of the article. From the results it is clear that a scientist wanting any really up to date patent information is going to have some considerable delay, particularly in the areas covered by the less efficient secondary services. Both the time lag in the inventor publishing his article in another form of literature, and then the time lag in this appearing in a secondary ser- vice have to be taken into account.

To sum up: we have found that the U.S. and Cana- dian literature is rarely duplicated by other forms of literature, and the coverage of this patent literature by the standard abstracting services leaves a lot to be de- sired. An information scientist should always, there- fore, make sure he searches an abstracting or indexing service which specialises in the patent literature if rea- sonable coverage of the literature of any scientific and technological literature is desired. Other research from this Centre (10) has made recommendations in the area of chemistry in this respect.

References (1) C. Oppenheim & E. A. Sutherland: Case Study on Galva-

lume 7 Journal of Chemical Information and Computer Sciinces 18: 126-9; 1878.

(2) F. Liebesny: Mainly on Patents, Butterworths, London 1972.

(3) R. T. Bottle & J. Singh: Unpublished results. (4) F. Liebesny et al: The scientific and technical informa-

tion contained in patent specifications - the extent and time factors of its publication in other forms of litera- ture - OSTI Report 5177, London 1973.

(5) C. V. Clark: Obsolescence of the patent literature - Jour- nal of Documentation 32: 32-52; 1973.

(6) Annual Report of the Department of Corporate and Con- sumer Affairs, Fiscal Year ending 31.3.68. Ottawa, Cana- da 1968.

(7) D. Kaye: Some shortcomings of Chemical Abstracts - North Western Newsletter (84): 5-7; 1965.

(8) C. Oppenheim: The patents coverage of Chemical Ab- stracts - The Information Scientists 10: 133-8; 1974.

(9) C. Oppenheim & D. J. Perhick: Unpublished results. (10) J. E. Freeman & C. Oppenheim: The patents coverage of

some chemical abstracting services - The Information Scientist 12: 83-96; 1978.

(11) U.S. Department of Commerce: Technology Assessment and Forecast, Eighth Report, USA 1977, 23-27.

80 World Patent Information 1 (1979) No. 2 Oppenheim - U.S. and Canadian Patent Literature