navigating the patent maze

When you have found a patent document of interest and want to know in which other countries it may have been filed or granted, the information is not always that easy to come by. Short of searching individual patent office databases (assuming that the patent office has one on the internet and in a language you understand), most searchers use the INPADOC database.

INPADOC collates information from over 70 patent offices around the world. The information is mainly bibliometric (e.g., name of inventor, name of assignee or applicant, priority documents, file number), but legal information (e.g., has the patent gone abandoned) is provided from some about 40 of the patent offices.

Until now, there have been very few free or low-cost sites for obtaining the information. And even the pay sites typically do a very poor job at displaying the information in a form that can be comprehended. Now however, there are two free sites, and both do a credible job at display. Check them out:

IP Newsflash: search page. This site is easier to negotiate.
National Informatics Centre in India: search page.

The cambiaIP website has long had complete text searchable patent documents in the field of agriculture. This week, the dataset is expanded to include all documents in life sciences and chemistry. The documents span the last 20-25 years and originate from the United States, Europe (granted patents only), Australia (granted patents only and from 1998), and WIPO.

Unlike most other free patent search sites that only allow text searching on the front page of a document, the cambiaIP site offers true full-text search, meaning that the entire text of the patent document can be searched. Moreover, multiple search terms can be limited by nearness, that is ANIMAL NEAR/10 DOG requires that the two search words, animal and dog, be located within 10 words of each other. This powerful feature can weed out many irrelevant patent documents.

To search the datasets, go here.
To explore other patent information on the site, go here.

The link leads to a PDF file of a powerpoint presentation regarding searching DNA and protein sequences, including those from patents, on the web. Although the presentation is biased toward STN International (the presenter is a sales manager from STN), it contains some interesting facts and useful information.

“The Complete Biosequence Patent Search: Mission Impossible?“, slides from a talk by Rob Austin from FIZ Karlsruhe at the Spring ACS National Meeting, 2004.

A reader has kindly informed me that the on-line database of Indian Patents is not provided by the Indian Patent Office. Rather, the database is provided by the Patent Facilitating Centre (PFC).

PFC was set up in 1995 by Department of Science and Technology under the auspices of the autonomous organisation Technology Information Forecasting and Assessment Council (TIFAC).

From the United States Patent and Trademark Office:

The report presents a preliminary list of the U.S. universities receiving the most patents for invention (i.e., utility patents) during the 2003 calendar year. All campuses are included.

1-University of California- 439 patents
2-California Institute of Technology-139 patents
3-Massachusetts Institute of Technology-127 patents
4-University of Texas-96 patents
5-Stanford University-85 patents
6-University of Wisconsin-84 patents
7-Johns Hopkins University-70 patents
8-University of Michigan-63 patents
9-Columbia University-61 patents
10-Cornell University and University of Florida-59 patents each

University names can be some of the most problemmatic for searching. Let’s say that you want to find all 439 patents issued in 2003 that were assigned to the University of California. First off, you can’t limit searches for a particular year on the patent search form or on the assignment search form. Even if you could limit for the year 2003, a search for “university AND california” in the Assignee field will not return all 439 hits. Why is that?

The reason is three-fold. (1) There may be typographical errors of the assignee names, (2) there are different equivalent names, and (3) the assignee has changed since the patent issued. Number 3 won’t be discussed further (but see the post on the Assignment database at the USPTO). Some examples of reasons 1 and 2 will illuminate the problems.

Using the Corporate Tree database at Delphion, a fee-based patent database site, a search for the assignee “university of california” resulted in a list of the different names found in the USPTO assignment data. The proper and correct name is “The Regents of the University of California”. Some of the typos (typos are underlined) include:

“The Reents of the University of California”
“The Regents fo the Univerisity of California”
“The Regents of the University of Calfornia“

Some of the equivalent names include:

“UC Regents”
“The Regents of the Univ. of California”
“The Regents of the University of Calif.”
“Univ. of CA”

What should be obvious is that finding all patents, or patent applications, assigned to a particular entity can be challenging. Sometimes even extremely challenging. Furthermore, words in other searchable fields, such as the inventor name, title of the invention, etc., may also have typographical errors or alternative spellings (e.g., color vs colour; economize vs economize). Unfortunately, there is no easy solution to these issues.

The data for United States published patent applications pose challenges for searchers who want to find applications filed by a particular assignee — e.g. company, university. Unlike the rest of the world, an applicant for a U.S. patent must be the inventor(s). Most of the time however, the inventor assigns rights in the invention to an organization, typically his employer. The assignment, which is a contract between the inventor and organization, is recorded at the Patent Office and appears on the cover page of any ensuing patent. The paperwork that is filed with a patent application asks for the name(s) of the applicant(s) — the inventors. Assignments are separately filed at the Patent Office.

A patent application is published 18 months after it is filed. The assignee data may not be available at the time of publication. If it is not available, only the inventors will be listed. Therefore, a search for applications filed by a company or other organization will not necessarily yield a complete set of results. Conversely, an application that is uncovered by a search based on other criteria will not necessarily inform the user who really owns the invention.

What can be done to overcome this problem? One trick I use is to do a subsequent search of U.S. patents and WO applications using the inventors’ names as search terms and see who they have assigned to recently. If the patent dataset doesn’t give any useful results, then try searching a scientific database like PubMed. The affiliation of the authors is almost always listed.

[See also post on "more about searching for assignees"]

Yesterday, the Cambia IP Resource implemented Dekko version 3 for searching patent documents. The user will see a few cosmetic changes on the search interface page, but will find that the major change is a faster response time in returning search results. The new Dekko software is the first of the improvements on the route to expansion of the patent datasets, which is planned for completion the end of March 2004.