We developed a scheme for classifying utility patents into technology
categories and subcategories. We began by attempting both broad and narrower
definitions of "agricultural biotechnology" and "food biotechnology."
Agricultural biotechnology is biotechnology applied to agriculture.
Most generally, agricultural biotechnology is the use of organisms or
parts of an organism to make or improve products or processes in agriculture.
The domestication of plant species and selection of desired characteristics
within agricultural species would qualify under this definition. More
narrowly, what biotechnology represents today is new knowledge
about the natural processes of DNA replication, breakage, ligation, and
repair that has made possible a deeper understanding of the mechanics
of cell biology and the hereditary process itself (McCouch,
2001). Although in agriculture the term "biotechnology"
has been most closely associated with genetic manipulation at the DNA
level, or genetic engineering, it may refer to a variety of techniques
and products. These may include, for example, use of molecular markers
in genetic improvement or more general use of genomic information.
Similarly, the use of enzymes for fermentation in brewing or cheesemaking
would be early examples of a broad definition of "food biotechnology."
Genetically engineering yeast to modify or improve a baking process would
be an example of a narrower, or more recent definition of food biotechnology.
In the complex and constantly changing world of agricultural biotechnology,
definition and classification are difficult tasks. The U.S. Patent and
Trademark Office (USPTO) has its own classification system, with classes
and subclasses, but these are intended to encompass all utility patents.
"Agricultural Biotechnology" patents can be found in several
major USPTO categories or some of their subcategories, including 800:
Multicellular Living Organisms And Unmodified Parts Thereof And Related
Processes; 935: Biotechnology (a now-abandoned category); 435: Chemistry:
Molecular Biology and Microbiology; 47: Plant Husbandry; 424 and 514:
Drug, Bio-Affecting and Body Treating Compositions; and 504: Plant Protecting
and Regulating Compositions. Agricultural applications may cut across
more than one of these categories. Furthermore, USPTO categories and classifications
change with evolving technology. For example, the 935 category, while
useful in the filtering process, no longer exists and cannot serve as
a guide to complete classification.
General Principles for the Development of the Classification Scheme
Several general principles governed the development of the ERS classification
scheme. First, because defining biotechnology is difficult, we wanted
to create a very broad database. Users would then be able to choose patents
from different subsets of the data to conform to their own definitions
of "agricultural biotechnology." The broadest definition would
result in selecting all the utility patents in our database, although
of course the filtering process as well as conceptual difficulties undoubtedly
excluded patents relevant for some users. On the other hand, a focus on
genetic engineering in agricultural plants could be served by selecting
patents in several categories relating to genetically engineered plants,
transformation technologies, and so on.
Second, from the beginning we tried to distinguish biotechnology products
(e.g., new crop varieties genetically engineered to resist pests) from
research tools or enabling technologies (e.g. genetic transformation through
micro-projectile bombardment or selectable marker genes).
Third, as with the USPTO classification scheme or the international classification
scheme, categories were expected to overlap. For example, a patent might
be classified under "genomics" because it covers identification
of a beneficial plant genetic trait with genomic sequencing of its DNA;
under "genetic transformation" of a plant to include this trait;
and under "protection, nutrition, and biological control of plants
and animals" because the transformed plant expresses an improved
agronomic property such as higher yield, pest resistance, drought tolerance,
etc.
Creating the Classification Scheme
An Iterative Process
The classification scheme was first developed before we populated the
database through the filtering process. ERS
researchers began by drafting a classification scheme, based on consultation
with several major sources. After this, scientists with experience in
agricultural biotechnology reviewed the scheme, in particular one of our
IFAFS (Initiative for Future Agriculture and Food Systems) collaborators
at Rutgers University, a molecular biochemist, who reviewed the classification
system on several occasions. This resulted in additions and modifications
to the scheme initially proposed. This process proceeded through several
iterations.
Major Sources Consulted
In addition to the major input received from our agricultural biotechnology
research collaborator, four major sources contributed to the classification
scheme: (1) the USPTO classification scheme, with additional information
provided by personal communication with the biotechnology art group within
the USPTO; (2) the biotechnology classes used by the Derwent World Patents
Index, a private sector subscription service that provides information
on patents granted in more than 40 jurisdictions worldwide; (3) a classification
scheme developed by biotechnology specialists at ISNAR (International
Service for National Agricultural Research); and (4) a classification
system developed by University of California-Berkeley researchers, after
they had used proprietary software developed by Aurigin, Inc., to group
agricultural biotechnology patents based on use of similar terms.
Simplification and Consolidation
As noted, classifying agricultural biotechnology patents is complex. The
process described above at one point resulted in a scheme with 15 main
technology groupings and 206 sub-technologies. In an attempt to streamline
the scheme, make it more user-friendly for a web-based application, and
create a scheme in which most of the subtechnologies would be populated
with patents, we simplified the classification system by combining or
dropping subtechnologies, and combining main technology groups. This resulted
in the present system with 7 main technology headings and 58 subtechnologies
(see classification scheme).
Implementing and Testing the Classification Scheme
Two basic approaches to determining whether any individual patent might
fall into one or more technology categories were tested. The first would
be to examine each patent individually and categorize it. This would have
the advantage of eventually classifying every patent in the database,
as well as providing an indepth review of "problem" patents.
The disadvantage of this approach is the inordinate amount of time that
would be necessary to classify all patents in the database into multiple
categories. This disadvantage was borne out in a trial run in which we
manually classified patents into a single simple technology class.
The second approach would be to write a series of logical rules (i.e.,
concerning specific words in specific parts of the patent, or USPTO headings
and subheadings) that would determine whether a patent would fall into
a given category or not. The advantages of this approach are closely linked
to the disadvantages of the first approach, and its disadvantages closely
related to the advantages of the first approach. Classifying patents with
logical rules is considerably faster (although the rules first have to
be written and tested), but in many if not all cases, classification might
be less accurate. Some patents might be placed into categories in which
they did not belong; others might not be classified into patents in which
they did belong. In the end, we opted for this approach.
We developed a set of logical rules intended to sort patents into at least
one of the technology categories; multiple categories for a single patent
were allowed. We then applied these rules to the patents that were populating
the database at the time. We took samples of the patents sorted into each
major category, and checked the patents in the samples individually to
make sure they belonged in the category to which they were assigned. We
also took samples of patents that were not sorted to any category and
checked the patents in these samples to see which category they might
belong in. On the basis of these procedures, we then modified the logical
rules for each technological category. The logical rules may be modified
further as we see the impacts of applying the rules as they now stand.
For example, out of over 11,000 utility patents currently in our database,
about 1,600 currently are not sorted into any category. Obvious next steps
would be to modify the rules further so that very few patents in the database
remain uncategorized; to check the patents in a given category to make
sure they belong; and to apply the rules to patents that were not filtered
into the database originally to see if the database should be expanded.
Citation
McCouch, Susan R., 2001, "Is Biotechnology an
Answer?", in Who
Will Be Fed In The 21st Century?, edited by Keith Wiebe, Nicole
Ballenger, and Per Pinstrup-Andersen, International Food Policy Research
Institute (IFPRI), Economic Research Service of the U.S. Department of Agriculture,
and the American Agricultural Economics Association (AAEA).
|
