Similarity between TRIZ Principles
Editor | On 11, Sep 2005
By: He Cong, Loh Han Tong
Developed during the initial patent research by Altshuller and his colleagues, the 40 Inventive Principles (IP) are one of the most important tools of TRIZ. As a basic tool of TRIZ, the IP were always introduced as the first concept to new comers to TRIZ. With the increasingly popular application of TRIZ in recent decades, experts have summarized numerous examples about application of IP in different technical and non-technical fields. However, some criticisms have also been given upon the original list of Principles: they are too abstract and sometimes overlapped with each other. In recent research, the relationship between IP has been analyzed. In 1998, Williams analyzed the symmetry and asymmetry of IP and summarized several groups of Principles that are opposite to each other. In 2002, Mann proposed a 5*3 matrix to group most of the Principles into five different strategies by the space-time-interface entities.
In an automatic patent classification system based on 40 IP has been proposed and analyzed upon a small set of patent documents. Elaborating on the idea of automatic patent classification according to IP, this paper analyzes the similarity between Principles based on the text information in the descriptions of examples using the Principles. 6 of 40 Principles are defined as ‘Obscure Principles’, which are hard to be analyzed by automatic classification system. In addition, two kinds of similarities between Principles are defined in this paper: text similarity and meaning similarity.
1. Text analysis of IP
1.1 Obscure Principles vs. Distinct Principles
As mentioned earlier, our analysis of Principles is based on the text information used in the descriptions of inventions. It is necessary for automatic patent classification according to IP in order to be able to assign the invention correctly. During our research, we found that for some Principles (34 of them), there are obvious descriptive text information, which hints at the Principles used. These Principles are defined as “Distinct Principles” in this paper. For instance, if an invention uses Principle 25, “self-service”, its patent description usually mentions the text information such as “self-”.
However, for the other 6 Principles, there is little text information that points towards the Principles used and inventions using such Principles share few similar or common words. These Principles are defined as “obscure Principles” in this paper.
1.2 Text similarity
The descriptions of inventions using some Principles share superficially similar text information which is distinctive to identify a group of Principles but is hard for an automatic classification system to differentiate among the group of Principles. For example, “pre-shrunk jeans” uses Principle 9 (prior counteraction); “pre-deposited blade in a surgery cast facilitates removal” uses Principle 10 (prior action). The descriptions of both examples, like many typical examples using these two Principles, share similar text information like “pre”, which is a typical descriptive text information for both Principles 9 and 10. However, deeper understanding is needed to differentiate between Principle 9 and 10. Groups of Principles like this are defined as “Principles with Text Similarity” in our paper. Other groups of Principles with text similarity are listed as in Table 2.
As shown in Table 2, the same Principle might appear in different groups. E.g. Principle 7 has relationship of text similarity with Principle 31: when Principle 7b is used, i.e. “an object passes through a cavity of another object”, the description usually contains words like “cavity” or “pores” which are highly likely to appear in the description of a patent using Principle 31. However, another sub-Principle of Principle 7 is similar to Principle 30: when Principle 7a (an object is contained inside another one) is used, words like “inside”, “outside” or “wrap” are usually contained in the descriptions, which appears when Principle 30b (an object is isolated from outside environment) is used. Like Principle 7, most IP are subdivided into several sub-Principles to describe different ways the IP is implemented. Although these sub-Principles share a common macroscopic idea of the Principle, each sub-Principle has its own particular emphasis and the sub-Principles have a subtle difference with one another within the same broad IP. Therefore, the similarity between Principles is not necessarily transitive. For example, #7b is similar to #31; #7a is similar to #30. But #30 is not similar to #31. In total, 19 different Principles are contained in the 10 groups of Principles with text similarity.
1.3 Meaning Similarity
Some Principles partly overlap with others. E.g. Principle 25b, “make use of waste material and energy” is similar in nature to Principle 22, “convert harm into benefit”. E.g. the main idea of the patent example, “waste heat conversion system” (US patent number 6,450,283), is to use waste heat energy and convert useless or harmful energy to benefit. Both Principle 22 and 25b can be used. We define such Principles as “Principles with similar meaning”. Another group of Principles with meaning similarity is 13.b and 15.c.
In this paper, based on text description of inventions, we have divided 40 IPs into different categories. We have listed 6 ‘Obscure Principles’ and 34 ‘Distinct Principles’. Then similarity between ‘Distinct Principle’ is analyzed. In Table 1 and 2, we have summarized 10 groups of IP with text similarity and 2 groups of IP with meaning similarity. This work will contribute towards the possibility of automatically classifying patent documents according to IP: the documents using ‘Obscure Principles’ are hard to be identified using superficial text information, thus hard to be automatically classified; however, the documents using ‘Distinct Principles’ are usually described by clear and similar text information, thus they are relatively easier to be automatically classified. However, the similarity between IP makes classification between similar IP hard and IPs may need to be clustered into groups.
To achieve automatic classification, based on the work done here, it is possible to analyze the text description of an invention and broadly classify the invention into a group of possible IPs if the IPs involved are distinct ones. However, research needs to be done to realize how to do so to the level of individual IP as well as how to treat Obscure ones.
We would like to thank Mr. Teo Wei Kuan, Eddy for collecting and classifying patent documents.
 Terninko J, Zusman A, Zlotin B. Systematic Innovation: An Introduction to TRIZ: St.Lucie Press, 1998.
 40 invention Principles with examples. Available on:
 Technical and Non-technical examples available on:
 Mann D. Evolving the IP. The TRIZ Journal 2002; August Issues.
 Williams T, Domb E. Reversability of the 40 Principles of Problem Solving. The TRIZ Journal 1998; May Issues.
 Loh HT, He C, Shen LX. Automatic Classification of Patent Documents for TRIZ Users. (Submitted)
About the Authors
He Cong is currently pursuing PhD in Department of Mechanical Engineering from the National University of Singapore, focusing on text mining and TRIZ. She has received her Bachelor in Engineering from the Huazhong University of Science and Technology in China. She can be reached at firstname.lastname@example.org.
Loh Han Tong received his Bachelor in Engineering from the University of Adelaide, his Master of Engineering from the National University of Singapore (NUS) and his Master of Science and PhD in Mechanical Engineering from the University of Michigan. He is an Associate Professor and Deputy Head in the Department of Mechanical Engineering at NUS. He is also a Fellow of the Singapore-MIT Alliance, which is an innovative engineering education and research collaboration between MIT, NUS and the Nanyang Technological University, to promote global education and research in engineering. His research interests include data mining, rapid prototyping, robust design and computer aided design.