Multi-criterial decision making system and method
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06F-017/00
G06N-005/04
G06N-005/00
출원번호
US-0553126
(2006-10-26)
등록번호
US-7437343
(2008-10-14)
발명자
/ 주소
Josephson,John R.
Chandrasekaran,Balakrishnan
Carroll,Mark
Iyer,Naresh Sundaram
출원인 / 주소
The Ohio State University Research Foundation
대리인 / 주소
Standley Law Group LLP
인용정보
피인용 횟수 :
3인용 특허 :
10
초록▼
An architecture is disclosed for assistance with exploration of design and other decision spaces and for making decisions. These decision spaces may be very large. The architecture consists of three main components: A Seeker acquires candidates by generating or retrieving them, along with their scor
An architecture is disclosed for assistance with exploration of design and other decision spaces and for making decisions. These decision spaces may be very large. The architecture consists of three main components: A Seeker acquires candidates by generating or retrieving them, along with their scores according to one or more criteria. A Filter locates a relatively small number of promising candidates that are retained for further analysis. Various filters may be used to locate the promising candidates. A Viewer allows a user to examine trade-off diagrams, and other linked displays, that present the filtered candidates for evaluation, analysis, further exploration, and narrowing the choice set. The computational load of the Seeker may be distributed among a large number of clients in a client-server computing environment.
대표청구항▼
What is claimed is: 1. A computerized system for exploring a set of decision alternatives D1-Dn wherein each of said decision alternatives in said set is evaluated according to at least two criteria C1 and C2, said system comprising: (a) a first software component that produces a subset from said s
What is claimed is: 1. A computerized system for exploring a set of decision alternatives D1-Dn wherein each of said decision alternatives in said set is evaluated according to at least two criteria C1 and C2, said system comprising: (a) a first software component that produces a subset from said set of decision alternatives D1-Dn using a multi-criterial filter and values for said at least two criteria C1 and C2 wherein said multi-criterial filter produces said subset by: (i) comparing decision alternatives in said set of decision alternatives D1-Dn according to said values for said at least two criteria C1 and C2; (ii) removing from said set of decision alternatives D1-Dn decisions alternatives that are inferior according to said values for said at least two criteria C1 and C2 wherein inferiority is determined according to whether larger values for Ci or smaller values for Ci are preferred; and (iii) retaining in said set of decision alternatives D1-Dn only decisions alternatives that are not inferior to any other decision alternatives according to said values for said at least two criteria C1 and C2 wherein superiority is determined according to whether larger values for Ci or smaller values for Ci are preferred; and (b) a second software component that presents on a computer display said subset of said set of decision alternatives D1-Dn produced by said multi-criterial filter of said first software component, wherein additional decision alternatives in said subset of said set of decision alternatives D1-Dn are removed from said subset of decision alternatives, forming a favored sub-subset of decision alternatives, in consideration of specific values of said decision alternatives according to computed criteria applied to decision alternatives in said subset of decision alternatives, including said at least two criteria C1 and C2 used in filtering said decision alternatives D1-Dn. 2. The computerized system of claim 1 further comprising a third software component for obtaining said set of decision alternatives from a database. 3. The computerized system of claim 1 further comprising a third software component that uses a functional and compositional modeling language for producing said set of decision alternatives and for evaluating said decision alternatives according to two or more criteria by simulations of behaviors of said decision alternatives. 4. The computerized system of claim 1 wherein said multi-criterial filter of said first software component is selected from the group consisting of classical filters, toleranced filters, strict filters, superstrict filters, selective superstrict filters, discernable difference toleranced filters, two pass toleranced filters, and onionskin filters. 5. The computerized system of claim 1 wherein said second software component displays said subset of said set of decision alternatives D1-Dn in scatterplots with axes corresponding to said at least two criteria C1 and C2 such that decision alternatives selected within a first scatterplot are distinguished from other decision alternatives in at least one other display of said decision alternatives. 6. The computerized system of claim 1 wherein said second software component presents said subset of said set of decision alternatives produced by said multi-criterial filter in a multi-attribute display comprising a one-dimensional scatterplot for each of said at least two criteria C1 and C2. 7. A computerized method for exploring a set of evaluated decision alternatives D1-Dn wherein each of said decision alternatives in said set is evaluated according to at least two criteria C1 and C2, said method comprising: (a) producing a subset from said set of decision alternatives D1-Dn by applying a multi-criterial filter to said set of decision alternatives using values for said at least two criteria C1 and C2 wherein said multi-criterial filter produces said subset by: (i) comparing said values for said at least two criteria C1 and C2 for two decision alternatives Da and Db; (ii) removing Db from said set of decision alternatives D1-Dn based on application of a dominance relation ε to said values for said at least two criteria C1 and C2; and (iii) repeating steps (i) and (ii) for said set of decision alternatives D1-Dn until no additional decision alternatives are removed by application of steps (i) and (ii); and (b) displaying on a computer graphical representations of said decision alternatives in said subset of decision alternatives according to said at least two criteria C1 and C2. 8. The computerized method of claim 7 wherein said dominance relation is a classical dominance relation, ε is zero, and Db is removed if C1(Da) is superior or equal to C1(Db) and C2(Da) is superior or equal to C2(Db), wherein superiority for each criterion is determined according to whether larger values for C or smaller values for C are preferred. 9. The computerized method of claim 7 wherein said dominance relation is a toleranced dominance relation, ε has a non-zero value, and Db is removed if C1(Da) is superior to C1(Db) by at least ε, wherein superiority for each criterion is determined according to whether larger values for C or smaller values for C are preferred. 10. The computerized method of claim 7 wherein said dominance relation is a strict dominance relation, ε has a non-zero value, and Db is removed if C1(Da) is superior to C1(Db) by more than ε, wherein superiority for each criterion is determined according to whether larger values for C or smaller values for C are preferred. 11. The computerized method of claim 7 wherein said dominance relation is a super strict dominance relation, ε has a non-zero value, and Db is removed if C1(Da) is superior to C1(Db) by more than ε and if C2(Da) is superior to C2(Db) by more than ε, wherein superiority for each criterion is determined according to whether larger values for C or smaller values for C are preferred. 12. The computerized method of claim 7 wherein said dominance relation is a selective super strict dominance relation, ε has a non-zero value, and Db is removed if C1(Da) is superior to C1(Db) by more than ε and if C2(Da) is not ε-worse than C2(Db), wherein superiority for each criterion is determined according to whether larger values for C or smaller values for C are preferred. 13. The computerized method of claim 7 wherein displaying on a computer graphical representations of said decision alternatives in said subset of decision alternatives according to said at least two criteria C1 and C2 comprises creating a scatterplot wherein said axes of said scatterplot correspond to said at least two criteria C1 and C2. 14. A computerized method for exploring decision alternatives, comprising: (a) producing a set of evaluated decision alternatives D1-Dn by: (i) acquiring a plurality of decision alternatives; and (ii) evaluating each of said plurality of decision alternatives according to a plurality of criteria C1-Cm to produce said set of evaluated decision alternatives D1-Dn; and; (b) producing a subset from said set of evaluated decision alternatives D1-Dn by applying a multi-criterial filter to values for said plurality of criteria C1-Cm wherein said multi-criterial filter produces said subset by: (i) comparing said values for said plurality of criteria C1-Cm for two decision alternatives Da and Db; (ii) removing Db from said set of decision alternatives D1-Dn based on application of a dominance relation ε to said values for said said plurality of criteria C1-Cm; and (iii) repeating steps (i) and (ii) for said set of evaluated decision alternatives D1-Dn until no additional evaluated decision alternatives are removed by application of steps (i) and (ii) and remaining evaluated decision alternatives are trade-offs with respect to each other; and (c) presenting on a computer display graphical representations of said evaluated decision alternatives D1-Dn wherein each evaluated decision alternative on said computer display is from said subset of evaluated decision alternatives containing only evaluated decision alternatives that are trade-offs with respect to each other. 15. The computerized method of claim 14 wherein said dominance relation is a classical dominance relation, ε is zero, Db is removed if Ci(Da) is superior or equal to Ci(Db) for every criterion Ci in C1-Cm, wherein superiority for each criterion is determined according to whether larger values for C or smaller values for C are preferred. 16. The computerized method of claim 14 wherein said dominance relation is a toleranced dominance relation, ε has a non-zero value, and Db is removed if Ci(Da) is superior to Ci(Db) by at least ε for every criterion Ci in C1-Cm, wherein superiority for each criterion C in C1-Cm is determined according to whether larger values for C or smaller values for C are preferred. 17. The computerized method of claim 14 wherein said dominance relation is a strict dominance relation, ε has a non-zero value, and Db is removed if Ci(Da) is superior to Ci(Db) by more than ε for at least one criterion Ci in C1-Cm, wherein superiority for a criterion C in C1-Cm is determined according to whether larger values for C or smaller values for C are preferred. 18. The computerized method of claim 14 wherein said dominance relation is a super strict dominance relation, ε has a non-zero value, and Db is removed if Ci(Da) is superior to Ci(Db) by more than ε for every criterion Ci in C1-Cm, wherein superiority for each criterion C in C1-Cm is determined according to whether larger values for C or smaller values for C are preferred. 19. The computerized method of claim 7 wherein repeating steps (i) and (ii) for said set of decision alternatives D1-Dn until no additional decision alternatives are removed by application of steps (i) and (ii) further comprises changing tolerances ε in determining superiority on subsequent applications of steps (ii). 20. The computerized method of claim 14 wherein presenting on a computer display graphical representations of said subset of evaluated decision alternatives D1-Dn comprises presenting said subset of evaluated decision alternatives in scatterplots wherein axes of said scatterplots correspond to said criteria C1-Cm.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (10)
Yavnai Arie,ILX, Autonomous command and control unit for mobile platform.
Ferkinhoff David J. (New Bedford MA) Gong Kai F. (Pawtucket RI) Keay Kathleen D. (Fairhaven MA) Nardone Steven C. (Narragansett RI), Expert system for assessing accuracy of models of physical phenomena and for selecting alternate models in the presence.
Becker Timothy Leo ; Heinlein Philip David ; Martin ; Jr. James Allen ; Orth Richard Daniel, Integration of groupware with quality function deployment methodology via facilitated work sessions.
Amado Carlos Armando (444 Brickell Avenue #51-111 Miami FL 33131-2400), Method and apparatus for applying if-then-else rules to data sets in a relational data base and generating from the resu.
Fishkin Kenneth P. ; Stone Maureen C. ; Mackinlay Jock D., Method and apparatus for interactive database queries via movable viewing operation regions.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.