The IBM Patent Server
by Thibault Fayette
1. INTRODUCTION:
Even before taking Professor Cavicchi’s class on IP research tools at the Franlin Pierce Law Center, I had heard about the possibility to make a "free patent search" using two web sites on the Internet, which are the United States Patent Office web site and the IBM patent server web site (hereafter, "IBM PS"). However, at that time I cannot say that I was really interested nor sensibilized with prior art search or more generally patent search.
As a student at the Law Center, I had a great opportunity to use some of the most powerful on-line research tools commonly used in the field of intellectual property by professional searchers. During the spring semester of 1998, I had a valuable chance to contemplate many interesting IP research issues with guest lecturers from companies such as Questel-Orbit, Dialog and Lexis-Nexis and also with professional searchers who use their services. It did not take long until I realized that there was a common "taboo topic" during these lectures, namely the IBM PS. I mostly disagreed with the strong criticism of the guest lecturers against the IBM PS. I found it necessary to explore more about the IBM PS to rebut their criticism, and thus I chose the IBM PS as a topic for my research project after consulting Prof. Cavicchi.
In this paper, I will explain to non-professional searchers what are the of using the IBM PS and how to use it. I will also address the professional concerns such as confidentiality, and try to present my opinion as a "disinterested party" which may or may not be shared by the future readers of this paper.
2. ACCESS, TOOLS AND RESULTS WITH THE IBM PATENT SERVER.
2.1 The home page access
The IBM Patent Server is FREELY available to any person having an access to the World Wide Web and using any common browser.
The home page address of that web site is simply as follows:
Once connected, you are informed that "IBM lets you access over 26 years of U.S. Patent & Trademark Office (USPTO) patent descriptions as well as the last 23 years of images. The first entries date back to January 5, 1971." (As far as I researched, the oldest patent that is available on the IBM PS is No. 3557648 and was issued on January 26, 1971).
This page has a link called What’s new which informs you of the current coverage date available on the database and also the recent events affecting the web site like for example, the date of the last update.
I highly think of IBM’s bona fide efforts to make you aware that some information regarding patents might be missing and to give you a link to a list of missing or incomplete data. where you can see which patents are completely missing (i.e. patents issued in the week starting Feb. 18, 1992 and Dec. 10, 1985) as well as the number of images missing which I will discuss further below.
You also have a link to a FAQ (Frequently Asked Questions) page, as well as information on the "interesting" history and background of this web site and also a patent Resource Page.
2.2 Tools
With no doubt, the most interesting portion of that first page is the search box which enables you to make a quick search immediately and instinctively using keywords. In my opinion however, the most interesting tool offered by this web site is the alternate searches portion which links you to pages where you areable to make Patent Number Search, Boolean Text Search or Advanced Text Search.
AGENT:"Blank"
AND
ASSIGNEE:Bausch Lomb
I used the double quote to be sure that the server will only search for the exact word I entered and thus prevent stemming. For instance a search without the double quote would also have looked (if possible) for Blanke. It is notable that the Boolean Text Search is useful if you plan to use OR or AND NOT operators, while a search with the AND operator is possible "(by default) with the Advanced Text Search page.
It is regrettable that combining more than two fields is not possible and also that the search fields are limited to 6 categories. It would be also convenient if the user could make a search using the issuance date or the filing date of a patent. However, I actually found out that a search using more than two fields is possible with the Advanced Text search to make a "boolean type query" in more than two fields. In addition, I will also show how to make a class search although there is no specific field for the classification.
Another weak point of the IBM PS is that you cannot perform a full text search. In other words a search will be performed in the designated fields (Inventor, Assignee, Title, Abstract, Claims and Agent) plus some "extra" (like class) and not in the all disclosure of the patent. Therefore, by using the IBM PS, you might be missing some relevant prior art because the information was "hidden" in the specification of a patent. In addition to using operators such as AND, OR and NOT you may also consider using more "powerful" operators as listed below:
<ACCRUE> search for documents with the word following the operator. The more words will be found in the specified field of a patent, the better the score will be for that patent
<IN> search for a word in a specific field of the patent index.
The available fields are:
title -- for the title of the invention
abstract -- for the abstract
assignee -- for the assignee names
inventor -- for the inventor names
agent -- for the agent/attorney/firm
claims -- for the claims
otherrefs -- for the other references
usrefs -- for the U.S. references
names -- combines both assignee and inventor
summary -- combines both title and abstract
for example: engine <in> (title, abstract, claims)
<WORD> search for the exact word; while not performing any stemming. Equivalent to double quote.
<THESAURUS> search for a specified word as well as synonyms which is a very convenient and powerful tool.
You can use wildcards like * and ?.
* specifies zero or more alphanumeric characters. For example fire* will match fire, fires, fireman, firefighter, etc.
? specifies one alphanumeric character. For example ?am will match cam, ram, jam, etc.
You can also use proximity operators.
<NEAR> search for two or more words within n number of words of each other. n has to be an integer between 1 and 1024. <NEAR> alone defaults to <NEAR/256>
<PARAGRAPH> is equivalent to <NEAR/100>
<SENTENCE> is equivalent to <NEAR/15>
Now, here is a "tip" that I found useful in this Advanced Text Search page.
Let’s assume that you are looking for all patents having the word "polymer" in the title, with M. Beck as an inventor and BASF as assignee. You just have to enter in the first box the following query:
(polymer <in> title) <and> (Beck <in> inventor) <and> (BASF <in> assignee)
6 patents satisfying the queries came out as a result.
I could have used other logical operators instead of <and> like <or> or <and> <not>. Thus it is possible to make a boolean search in more than two fields on the Advanced Text search page.
2.3 Results.
When you launch a query on the IBM PS, usually the results can be obtained fast. However, since you are using a database accessible through the Internet and that you are not alone on earth "surfing" on the Net, the speed can be sometimes very slow.. Therefore, IBM cannot be blamed if it sometimes takes a lot of time to have the results. Nevertheless, I find the access time to the server very reasonable in general.
Let’s assume that you are looking for all the patents granted to a French corporation called Renault. Just type Renault in the assignee box and click on "search".
After a few seconds, the server provides you a list of a maximum of 200 patents with their titles that match your query. These patents are organized and listed according to "the score of the hit" and according to the issuance date of the patent.
In this example, the IBM PS had to look for the word "Renault" in the assignee field and thus gave a better score to "hits" where that word appeared the more often. Accordingly the first patent given in the results, although issued before the subsequent patents which appeared on the list, had the first rank since the word Renault appeared twice in the assignee field instead of only once for the other patents.
If you want to access the patent, you just have to click on its number and the server will lead you to its first page.
Typically, the following information will be provided :
Patent number with the title, Inventors, Assignees, Issuance and filing date of the patent, serial number of the application, Main Status of the patent, International class, U.S class.
You also have the abstract, a list of all the patents cited by your patent (each of them accessible by clicking) and a link to all the patents which cite your patent.
The server provides you the first claim of your patent and a link enabling you to read all the claims. The server informs you if there is any U.S application related to this patent as well as the foreign patents cited or used as priority for filing and also if other kind of references have been cited during the prosecution. You can also see who was the attorney, agent or law firm who prosecuted the application and who were the examiners at the U.S.P.T.O.
As discussed above, although there is no such field box for class number, you can anyway make a class search by the two following ways.
The easiest one is as follows. Let’s assume that you already know a patent having the class you want to search for, just search for that patent and then, click on the specific class you are interested in. The server then gives you ALL (without being limited to 200 hits) the patents belonging to that class.
Second, if you do not know any patent but know the class and subclass your are looking for, just type it in the first box of the Advanced Text Search: e.g. , 180/197
The server will look in all the fields including U.S and International class and then gives you a list of patents. The probability that you can find patents with the class is very high. Once you have at least one patent with the class you are looking for, by clicking on it you will have a "clean" result. You can also do the same thing with the international class. My last advise is that if you are looking for a class made of only two digits (e.g. 45/123), put a 0 before these two digits ( 045/123). That way you will narrow the number of undesired patents.
The Maintenance Status field indicates if the USPTO has reported that this patent has lapsed due to unpaid maintenance fees or other special conditions. That fields informs you of the actual status of the patent you are looking for.
The followings are fourteen possible values:
The default (no value) status is that the USPTO has not reported the patent as having lapsed.
CC: Certificate of Correction issued which only indicates Certificates of Correction issued since 1986.
E1: Expired four(4) years after the initial issue date.
E2: Expired eight(8) years after the initial issue date.
E3: Expired twelve(12) years after the initial issue date.
R1: Reinstated after E1
R2: Reinstated after E2
R3: Reinstated after E3
B1: Re-examined
B2: Re-examined 2nd time
B3: Re-examined 3rd time
W: Withdrawn. Patent number was assigned and later withdrawn.
XT: Term has been extended
M: Missing from this database, but believed to exist. Data will be added to the entry as it becomes available.
In my opinion, it was a very good idea of IBM to add a link to its Licensing Department page in all the patent that IBM is willing to grant a license.
For example, you can have a look to the patent numbered 5644432.
Although there might be some difficulties, I believe it would add more values to the server if willful patent owners advertised their intent to grant licenses by using the IBM PS.
Last but not least, you have a free access to all the images of the patent just by clicking on the link "view images". After you click, a new window is opened, which let the possibility to do something else in the same time, and your computer fetches the first image of the patent.
You can enlarge, reduce, print or even save the image in a GIF format which is inferior in quality to the TIFF format. The best way to print the patent is first to save it in your hard drive and then to open it using for instance a word processor and then print it. This feature is very useful because it enables you to have an access to drawings which are critical in patent search, especially in the field of mechanical engineering.
If you are not satisfied with the relatively mediocre quality of the images of your patent you still have the option to order the patent by using the form provided via Optipat for $ 2.50.
3. REAL CONCERNS OR PARANOIA ABOUT THE IBM PATENT SERVER
As mentioned at the beginning, a lot of people are concerned or even "scared" of using the IBM PS. Is Big Brother watching at you?!
We should consider the interest of a third party "spying" your communications between your computer and the IBM PS in order to gain an advantage but we should also consider the potential interest of IBM for your queries.
Here is the problem.
A few sentences in the "Conditions of Use" page of the IBM PS are somehow confusing to the reader.
"The communications between IBM and the users of the IBM Patent Server site are not of a confidential nature. Therefore, any communications that you wish to remain confidential, whether they are search queries, guestbook entries, e.mail, or any other communications, should not be relayed to the Patent Server."
...
"IBM does not maintain records of the search entries submitted by, or results obtained by, individual, identifiable users of the IBM Patent Server."
The first sentence can be interpreted in several ways and could be almost threatening because of the "not of confidential nature" part. It seems to me that IBM simply wants the potential users of the server to be aware that the transmissions between the user’s computer and the IBM PS using the Internet, are not encrypted (like most of the transmissions on the Net) and thus are potentially "interceptable" by others.
The word "potentially" is incredibly important and that is the reason why before starting to use the server, you should reasonably assess the probability that somebody out there is watching carefully to all your communications through the Internet. However, there is no reason to become completely paranoid. I would simply say that if you are working in or for a very High-Tech corporation and/or involved in a very competitive field, you should better not use the IBM PS, just in case!
Regarding the second sentence, it is obvious the IBM has the capability to keep records of your search or queries when you are using their server. Moreover, IBM solely admits that it does not maintain records... obtained by individual, identifiable users. What does this sentence mean ?
It means for example, that when I am using the IBM PS from the Franklin Pierce Law Center which domain name is piercelaw.edu, IBM might know that someone from fplc is using the server.
It could be annoying for an employee whose corporation has its own domain name, but there is an easy way to avoid that problem. You can just contract with a private Internet access provider to be granted an "anonymous" (for IBM) account and thus never disclose your corporation’s domain name.
One of the main concerns of the users is that their query and results are not to be confidential and IBM might be able to use their search against their interest I find this argument very implausible and a few examples will better explain my rebuttal than a long reasoning.
Before filing an application for a utility patent, you decide to make a "light" prior art search using the IBM PS. You enter several queries and edit several relevant patents.
Let’s assume that IBM or a third party knows exactly what you were looking for.
What could they do with that information ?
First of all, all depends on the field of research. I sincerely doubt that IBM is going to analyze carefully all the searches performed in the field of chemistry or mechanical engineering. But let’s assume anyway that your search has been performed in one of the fields that interests IBM.
IBM has your queries and the results of your searches which may or may not be relevant to your invention.
I can hardly imagine a case in which IBM would use a team of engineers or researchers in order to analyze your search and try to determine if a new invention or an improvement is hidden among these information. Certainly IBM could try to do it. However, how many searches could be so important that IBM finds it worth to investigate ?
Once again, before filing an application you make a prior art search using the IBM PS, and IBM or a third party knows the results of your search. You found 13 relevant patents and that one of those is seriously annoying because it potentially constitutes a statutory bar for your patent. During the prosecution, you disclose only 12 of the patents found and you pretend to forget (in fact breach your duty of candor and deal in bad faith with the PTO) the 13th. You are incredibly lucky and the examiner at the PTO does not find that prior art and issues your patent. After a few years, you discover that somebody (the third "spying" party) is infringing your patent and thus you decide to bring a suit.
Could that third party, try to invalidate your patent by providing to the Court the 13th patent you never disclosed (as required since you knew about it) and also try to show that you breached your duty of candor in dealing with the PTO ?
Potentially the answer will be YES.
However concretely the probability it would happen is e , i.e. closed to 0.
Then, if the IBM PS is so safe with no risk at all and if IBM does not intent to take advantage of you searches, you may probably wonder why IBM provides these services for free ?
First, I believe that it should not have IBM much to provide such a service and to maintain such a site.
The data is probably furnished by Optipat in the GIF format in consideration of soliciting potential customers who may order a paper patent from Optipat.
Moreover IBM has a tremendous potential in using its own hardware (server), software, database, Internet access provider (IBM Global Network), and all these assets must have made it relatively easier to enable this service.
IBM introduces the technology used to maintain this site.
"This site uses IBM's Internet Connection Server to serve all patent information pages to web clients. IBM's Net.Data cgi program serves as the web gateway to the DB2 Universal Database, which retrieves from our 24 Gigabyte database of patent text information. Patent Server Guestbook and FAQ pages are served via Lotus Domino webservers. All systems are running on IBM RS/6000 and SP servers using the AIX operating system. The multiple server nodes are robustly combined using Interactive Network Dispatcher. The content is delivered to the Internet via the IBM Global Network.
In addition to this central site, IBM has developed a stand-alone Patent Library Site Server, which allows users to get this site's capabilities in a self-contained machine."
This is once again my own opinion but this web site can function as a very good show case for IBM where IBM can advertise its know how and technologies to potential customers. IBM also offers to sell the equivalent of the web site to work as a patent library in an Intranet network.
This IBM PS might evolve in the future.
"Possible enhancements include expansion of world wide patent coverage working with other organizations."
If IBM seriously plans to provide information regarding the European Patent Office and Japanese Patent Office, The existing patent data providers may have to face a threat to their flourishing businesses and thus they may want to "try" discreetly to discredit the IBM PS.
4. CONCLUSION.
The IBM PS is certainly not a perfect tool to make a prior art search (missing data, not a full text search, confidentiality issues). However it deserves anyway some respect for making accessible to a large public, public information for not a penny. Some law firms, some companies and some individuals should seriously make a cost-benefit analysis integrating the confidentiality issue.
Some concerns of the professionals are certainly but as one of my professor likes to say "this is a free country", thus you are completely free to prefer on-line service providers and to pay for the same information with certainly a better service.
I would like to thank Prof. Cavicchi for his precious help for this project and also Mrs Nancy Lambert for her article "But what’s in it for IBM?: Patents on the Net." Searcher 5(8), Sept 1997, p. 33-37