Data processing: artificial intelligence – Knowledge processing system – Knowledge representation and reasoning technique
Reexamination Certificate
2000-12-13
2004-09-21
Knight, Anthony (Department: 2121)
Data processing: artificial intelligence
Knowledge processing system
Knowledge representation and reasoning technique
Reexamination Certificate
active
06795815
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to a computerized knowledge mining, intelligence analysis, and knowledge management system, specifically to analytical logic (i.e. knowledge) capturing, constructing, accumulating, and sharing.
2. Related Art
Intelligent analysis process often involves a variety of data and reasoning logic. In order to make better decisions or analyses, people need to be able to utilize multi source knowledge. The biggest issue is how we utilize multi source knowledge understandably and efficiently. For example, when developing a new stock performance evaluation method, we may like to combine different analytical logic or methods such as technical analysis, fundamental analysis, chart analysis, volume analysis, and market psychological analysis, which raises some questions. Can people quickly build the method at their skill levels? How can people combine different analytical logic and/or methods together without losing logic clarity, flexibility, scalability, and share ability? Can the combined method be easily updated and extended? All of these questions boil down to a single issue: how do we capture or computerize human analytical logic that can be easily shared, especially across analysis areas and/or industries.
Although any analytical logic can be constructed by “CASE” or “IF . . . THEN . . . ELSE . . . ” programming statements, the readability, scalability, and changeability of the analytical logic often become barriers for sharing and updating knowledge when the analysis issue become more complicated.
We all agree that human analytical logic or reasoning processes can be well presented by a (decision or knowledge) tree structure. Because of the unique tree's characteristics such as independency of peer nodes and single parent node, the tree structure is a most scalable, flexible, and commonly used analytical structure. Although many decision-tree construction methods (e.g. Naïve-Bayes, Classification, Fuzzy, and Neuron Network.) have been developed, the structures of nodes are often not uniformed or standardized. Different decision-tree construction methods use different node structures. Even within the same construction method, sometimes, many different node structures (e.g. decision node, classifier node, data/factor node) are used. For a decision tree with multiple node structures, the analysis process, logic modification, and logic sharing (e.g. embed a decision tree into another decision tree that is built with a different construction method.) are often complicated, inflexible, and inefficient.
SUMMARY OF THE INVENTION
The present invention provides an open knowledge structure that defines a uniform knowledge node. Each knowledge node has a plurality of components and attributes. According to the present invention, the knowledge node includes a knowledge cell, a processing unit with a user specified evaluation function, a decision set, a user specified weight function, a learning matrix, and a user specified learning function. The values (i.e. basic knowledge or judgments) of knowledge cell can be static or dynamic. The static value can be a constant number or string. The dynamic value can come from an intelligent analysis function, a survey function, a data-mining tool, or an analysis module. A knowledge node can have zero to many inputs.
The present invention defines open structure and architecture for constructing dynamic and distributed intelligence analysis modules (i.e. knowledge trees). The knowledge tree can be stored in knowledge bases that are built with common used commercial databases (e.g. Oracle, DB2, Sybase, SQL Server and MS Access), data file, or executable file. According to the present invention, a knowledge tree has at least one knowledge node. The knowledge nodes are interlinked by using address-based link methods. The knowledge nodes or sub-trees of a knowledge tree can be stored in either identical or different knowledge bases that reside on either the same knowledge base server or different knowledge base servers.
The present invention introduces a method that can simplify the knowledge capturing processes, programming codes, and analysis processes. With the open and uniform knowledge structure and architecture, the analytical logic can be understood easily, shared across analysis areas and systems effectively, updated dynamically, and processed efficiently with peer-to-peer technology. According to the present invention, the knowledge trees can be effectively stored and managed in distributed manner. In summary, the present invention provides a knowledge-mining tool that enables people to construct their analytical logic using a variety of existing knowledge and/or methods.
Further features and advantages of the invention, as well as the structure and operation of various embodiments of the invention, are described in detail below with reference to the accompanying drawings.
REFERENCES:
patent: 4730259 (1988-03-01), Gallant
patent: 4860214 (1989-08-01), Matsuda et al.
patent: 4864490 (1989-09-01), Nomoto et al.
patent: 5131074 (1992-07-01), Nakamura et al.
patent: 5245698 (1993-09-01), Matsunaga
patent: 5251288 (1993-10-01), Nomura et al.
patent: 5347615 (1994-09-01), Yamakawa et al.
patent: 5398303 (1995-03-01), Tanaka
patent: 5432885 (1995-07-01), Nomoto et al.
patent: 5463718 (1995-10-01), Maeda et al.
patent: 5479567 (1995-12-01), Shinmura et al.
patent: 5586217 (1996-12-01), Ota et al.
patent: 5933818 (1999-08-01), Kasravi et al.
patent: 6012051 (2000-01-01), Sammon et al.
patent: 6182058 (2001-01-01), Kohavi
patent: 7-191852 (1995-07-01), None
A New Approach for Evolving Clusters, Robert E. Marmelstein and Gary B. Lamont, ACM 1998, pp. 268-274.*
Data Mining for Hypertext: A Tutorial Suvery, Soumen Chakrabarti, SCM SIGKDD Explorations Jan. 2000, pp. 1-11.
Holmes Michael B.
Knight Anthony
Zhang George Guonan
LandOfFree
Computer based knowledge system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Computer based knowledge system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Computer based knowledge system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3260815