Semyon D. Savransky
Abstract
The fund of world patents presents a unique source of deep technical knowledge. Yet, the patent fund itself represents only a potential source for acquiring knowledge. This potential cannot be expanded into deep knowledge without a formal procedure of patent study and an agent (human or computer) able to do this patent analysis. Here, for the first time, such a procedure is proposed and discussed in detail. The patent's peculiarities as a source of deep knowledge are outlined. Some elements of TRIZ contradictions are briefly described. The possibility of expanding the procedure to patent investigation by computers is discussed in terms of a data mining approach developed in the field of artificial intelligence. The role of patent analysis as the base of experimental TRIZ research is outlined.
KEYWORDS: contradiction, knowledge, patents, reusable information,
TRIZ
1. Introduction
There are several approaches to developing problem-solving techniques [1-7].
In the first approach, developers of problem-solving methods base their research on analysis of historical anecdotes and/or interviews with people who have successfully solved problems. This approach represents an attempt to extract the techniques of problem solvers in order to create a generic set of problem solving recommendations. Unfortunately, problem-solvers, designers, doctors, engineers and other white collar workers are often unable to articulate the nature of their process for resolving problems, hence the special (but not very success) questionnaires and games developed through artificial intelligence studies to recreate the methods of design that experts employ [3].
The second approach to structured problem-solving involves trying to define techniques based on the way the human brain is believed to work [5]. Efforts by cognitive science specialists have led to systems that collect and organize the knowledge of leading specialist in selected fields (technical or non-technical). These systems are quite useful in solving routine problems (especially problems that require numerous calculations and a great deal of input information), but are not yet successful in solving creative or inventive problems [3,6].
The third and final approach is the methodology developed by Genrich Saulovich Altshuller. This inventive problem-solving theory is based on the documentation of solved problems to enable the analogic search for successful solutions to unresolved inventive problems [1,2].
It is well known that reliable, well-organized knowledge helps solve problems [1-3]. From ancient times, shallow knowledge has been distinguished from deep knowledge. In terms of Altshuller's goal to develop a methodology for systematic resolution of technical problems, individuals, design teams, companies and even industries have shallow knowledge [1,2] with limited potential for resolving technical problems of a high difficulty [2,4].
The only last problem solving approach, named TRIZ, allows the effective resolution of difficult technical problems [4]. The basis for the strong advantage of TRIZ [1,2] in comparison with the other approaches [3,5] lies in Altshuller's choice of the right source for acquiring knowledge - the patent fund. Studies of the patent fund have enabled development of most of the modern TRIZ tools [1,2]. Nevertheless, most TRIZ specialists (or TRIZovszy in Russian) do NOT study patents. Perhaps this is simply because they do not know HOW. The reason for this may be that, as of yet, nobody has described a concise plan for patent study.
To that purpose, we will use the opportunity of this conference to:
2. Sources of technical knowledge
There are four major references of technical information:
Notes:
In contrast with all other sources, only patents:
Note:
The following URL are popular for patent searches:
http://patent.womplex.ibm.com/, http://www.uspto.gov/, http://patents.cnidr.org
3. Peruse of Patents
More than 20 millions patents have been issued world-wide, though there is some duplication between countries. Although about 85% of patents are not usable for TRIZ research [1,2,6], about 3 millions patents warrant rigorous and detailed studied. This study should be:
3.1 Formalities
In order to produce uniform patent study results that can be easily used by all TRIZ specialists, we need to establish some universal terms. First, we accept Altshuller's scale for division of problem solutions into five levels of inventiveness, as described in Appendix 1. Second, we define the following terminology and principles:
TRIZ+trust: the register of patents that represent strong solutions (Level 3 and above) that confirm TRIZ.
TRIZ-pool: the register of patents that represent solution that conflict with current TRIZ.
TRIZ Bank: the combined registers of the TRIZ+trust and the TRIZ-pool.
InfoBank: the register of patents that can be used as analogies and/or examples for new problems (usually solutions at Levels 3 and 2).
WasteDeposit: the register of non-inventive patents (Levels 1 and 2). For these patents, only legal information should be storage.
1. Completeness principles:
Patents fund = TRIZ Bank InfoBank WasteDeposit
TRIZ Bank = TRIZ+trust TRIZ-pool
2. Unique principle:
At any point in time, a patent can have only one entry in the patent fund.
Patent TRIZ-pool XOR TRIZ+trust XOR InfoBank XOR WasteDeposit
Entries are ordered in accordance with the level of curiosity of a particular patent search.
Note:
TRIZ Experts use the signs + and - as shorthand references for the TRIZ+trust and the TRIZ-pool, respectively. The InfoBank is auxilary to the TRIZ+trust, and is designated by 0 (the zero sign) while the sign X is used to designate the WasteDeposit.
The third requirement for establishing a uniform patent study procedure is having experimental patent study results reported using a standard form, shown in the Figure 1. Most terms used in Figure 1 are well-known to those familiar with the work of Altshuller [1]. Some terms are clarified by the author in earlier writings [2] (see also [5]).
The one of most important goals of patent study is the generation of
reusable information. The extracted information can then be transferred
into other databases, e.g. manual banks of technical effects [8] or computer
knowledge repositories [3,6].
|
00. Legal information Sign: +, -, 0 or X 0. Abstract of extracted knowledge that is new for TRIZ and keywords 1. State of the art (before the patent)
a) Contradiction a1) Principle of solution a2) Reduction rule b) Effect (Phenomena) b1) Natural b2) Technical c) Heuristics c1) Engineering Principle c2) Separation Principle c3) Standard Transformation c4) Prescript d) Trend of evolution e) Special cases
© S. D. Savransky 1981 /Rus./, 1997 /Engl./ |
Figure 1. Template for studied patent.
3.2 Basic Flowchart of Patent Investigation
Most issued patents are useless from the point of view of TRIZ developers [1,2,6]. Nevertheless, all patents can be investigated in the terms of traditional TRIZ [1, 2] (as shown in Figure 2), though Level 1 and 2 patents do not require a long period of analysis. These low-level patents are quickly scanned by TRIZ researchers and important legal information is noted in order to protect other researches from wasting their time. But conservation of time is also important when studying the higher level patents.
A small number of patents (Level 5 of the Altshuller scale) require
meticulous analysis within the framework of techno-economical trends of
noosphere development [9, 2]. A larger number of high-level patents (usually,
Levels 3 and 4 of the Altshuller scale) illustrate resolution of a problem
that had contained a contradiction. These patents require focused, critical
analysis that corresponds to the non-traditional description of contradiction
ontology proposed and developed by the author [2,6,10].
4. Computer Aid Study of Patents
The growing plethora of information in the worlds of science, business,
technology and government create a need for tools and techniques that can
analyze, summarize and extract "knowledge" from the raw data included in
huge and noisy databases of available facts. Most of the so-called data
mining tools and techniques are based on statistics, machine learning,
pattern recognition and artificial neural networks [3]. With some adaptation
[6], it seems that these tools can be used for knowledge discovery in the
patent fund. Hopefully, the computer-aided studies of patents will be faster
and, perhaps, more effective than the manual studies conducted until now.
5. Some results of patent studies
The author's first project after graduation of university in 1980 was quite unusual. The goal was to propose a solid state replacement of an electro-mechanical relay. Constrains for the proposed mechanism were minimal:
Figure 2. Schema for study of a single patent
During this project, the author studied about 5000 patents for different semiconductor, solid state devices and technologies. At the same time, the author was learning TRIZ through an extensive critical study of the Altshuller's methodology. Later, the author investigated about 1000 additional patents, mostly in the fields of semiconductors, fiber optics, new application of synthetic materials, and some advanced technologies.
Development of the new relay became secondary to the main result of this research: discovery of several new heuristics (rules of thumb) that can be used to represent the inventive principles, standard solutions, prescripts and other tools in the TRIZ framework [2].
All these studies were performed manually, although the author and some
students of the Virtual TRIZ College (http://www.trizexperts.net/VU-TRIZ.htm),
have used the web sites mentioned earlier to conduct governed patent searches
that confirm the established heuristics.
|
Extraction of a new heuristic /effect/ from patent(s)
TRIZ Bank or Fund of Selected Patents |
Figure 3. Part of circular flowchart for TRIZ research
6. Conclusion
The importance of the patent fund as a primary source of deep technical knowledge is confirmed by the fact that numerous attempts by psychologist, cognitive scientists and specialist in artificial intelligence to design problem-solving methodologies equal to TRIZ have, as yet, failed. Each science has its experimental, theoretical and application components (Figure 3). The experimental component of TRIZ can be based on a foundation of patents studies. Using the concrete knowledge provided by the information in patents, we can confidently state, "TRIZ is a science". This concrete knowledge is distilled through our systematic procedure of patents studies within the TRIZ framework. Fortunately, TRIZ has this source of deep knowledge. It is the responsibility of TRIZ specialists to use the patents fund for continued development of the methodology.
Acknowledge: I would like to thank Ms. Mary Ann Kahl for the technical assistance.
REFERENCES
[1] G. S. Altshuller, To Find an Idea: Introduction into the Theory of Inventive Problem Solving, Nauka, Novosibirsk 1991 (in Russian).
[2] S. D. Savransky, TRIZ, 1999, 454 pp. Forthcoming in English.
[3] S. J. Russell and P. Norvig, Artificial Intelligence : A Modern Approach, Prentice Hall, New York, 1995.
[4] S. D. Savransky, Attributes of the Inventive Problems , AAAI Spring Symposium on Search Techniques for Problem Solving under Uncertainty and Incomplete Information, Stanford University, March 1999.
[5a] A. B. VanGundy, Techniques of structured problem solving, Van Nostrand Reinhold Co., New York, 1988.
[5b] J. M. Higgins, 101 Creative Problem Solving Techniques : The Handbook of New Ideas for Business, New Management Pub Co., 1994.
[6] S. D. Savransky, How TRIZ and AI could enrich each other, AAAI-99 Sixteenth National Conference on Artificial Intelligence, Orlando - Florida, July 1999, Abstract ID: A375
[7] S. D. Savransky and C. Stephan, TRIZ: Methodology of Inventive Problem Solving, The Industrial Physicist, 1996, v. 2, No. 4, p. 22-26.
[8a] S. S. Litvin and A. L. Lyubomirskiy, About a Bank for Technical Effects, Journal TRIZ 1990 #2, p.22-27 (in Russian).
[8b] S. S. Litvin and B. Axelrod, Development Description. A Bank for Technical Effects, Journal TRIZ 1995 #1, p.55 (in Russian).
[9] V.I. Vernadsky, Scientific works, Nauka, Moscow 1989 (in Russian).
[10] S. D. Savransky, Human- and Technique-Like Contradictions in TRIZ, WWW TRIZ Journal 1997
[11] S. D. Savransky, Role of contradictions in problem solving, AAAI-99 Sixteenth National Conference on Artificial Intelligence, Orlando, July 1999, Abstract ID: A376
Appendix 1.
Altshuller's levels technical problem solutions.
Level 1: Regular. Routine design problems are solved (usually by trade-offs) using methods well known within the specialty or inside a company.
Level 2: Improvement. Existing systems are developed through qualitative but not substantial change. Development is usually due to application of non-common methods from the same industry, but with some additional knowledge from the inventor's specialization or/and some creative effort. This 'non-common' method can already be found in the TRIZ knowledge-base.
Level 3: Invention inside paradigm. An existing system experience essential improvement and radical change through utilization of methods or knowledge from other fields, sometimes far removed from the existing system's industry.
Level 4: Breakthrough outside paradigm. A new generation of a system is created using a solution that has been found in science rather than in technology.
Level 5: Discovery. Pioneer invention of a radically new system is usually based on a major discovery in some basic (or new) science.
The more inventive solutions correspond to the Levels 3 through 5.
More details can be found at the WWW site http://www.trizexperts.net
or in References 1,2, 7.
About the author:
Semyon D. Savransky, member of the International TRIZ Association, became acquainted with TRIZ in 1980-1981 through systematic study of Altshuller books in Russia. His encyclopedic knowledge of natural science and electronics was soon recognized in the TRIZ community. He has applied TRIZ for several R&D projects in various high-tech industries and for pure scientific research. Semyon D. Savransky holds eight patents, and is the author of more than 150 scientific papers concerning the development of TRIZ, in physics and materials science. Dr. Savransky received his Ph.D. in Leningrad (1989). His academic background is divided between Novgorod State University (Russia), University Pais Vasco (Spain) and New York City University (USA). Semyon D. Savransky is currently the head of a Division of the West Coast Quartz Corporation (California). He is also the founder of the Research Center at Novgorod State University - NGPI (Russia) and TRIZ Experts International Company.
Contact Information:
TRIZ Experts
6015 Pepper Tree Court Newark, CA 94560-4745, USA
URL: http://www.trizexperts.net
E-mail: TRIZExperts@hotmail.com