Classifying the TRIZ Contradiction Problem of the Patents Based on Engineering Parameters - Technologies and Applications of Artificial Intelligence - page 342

Classifying the TRIZ Contradiction Problem
of the Patents Based on Engineering Parameters

Chung-Kai Tseng, Chih-Heng Chung, and Bi-Ru Dai

National Taiwan University of Science and Technology,
#43, Sec.4, Keelung Rd., Taipei, 106, Taiwan, R.O.C
{M10015080,D9915015}@mail.ntust.edu.tw
[e-mail]

Abstract. TRIZ is a useful theory to solve the engineering contradiction prob-
lems. One of technological methods of TRIZ is Contradiction Matrix which is
widely used for the solution of technical contradiction problems. For the TRIZ
users, finding out some patent documents which had solved the same contradic-
tion is helpful to solve the problems of this type of contradiction. Classifying
patents contradiction based on the Engineering Parameters is more reasonable
than the Inventive Principles, but all existing patent classification researches are
unsuitable on patents contradiction classifying based on the Engineering Para-
meters directly. In this article, a new algorithm named MCIVC for classifying
patents technical contradiction based on Engineering Parameters is proposed.
This multi-layer classification algorithm adopts the associated rule-based ap-
proach combining the lazy learning. It does not only consider the semantic rela-
tionship among terms, but also consider the syntactic structure between words.

Keywords: patent classification, TRIZ, Contradiction Matrix, contradiction,
multi-layer classification, associated rule-based approach, lazy learning.

1 Introduction

TRIZ, the Theory of Inventive Problem Solving proposed by Altshuller has become a
more well-known and useful theory to solve the engineering contradiction problems.
Altshuller addressed that only 1% of patents are genuine originations and the rest are
based on known ideas and conceptions of the former plus novelty methods. He also
recognized that the basics of innovation and new ideas were not in the brains of
inventors, but in the published inventions [1,2,3]. He discovered that many problems
arose because of technical contradiction between Engineering Parameters. The tech-
nical contradictions are solved by the Contradiction Matrix that utilizing 39 Engineer-
ing Parameters and 40 Inventive Principles [3].
Patents have already become very important knowledge source for product develop-
ment, research, and innovation. Because of the tremendous amount of patents published
in a short time, automatic patent classification becomes a very popular research issue of
data mining. From the view of the Contradiction Matrix, most of contradiction occurs
when there is a conflict existing between two Engineering Parameters. However, past