UserPreference.ppt - Semantic Web Lab

Document Sample
UserPreference.ppt - Semantic Web Lab Powered By Docstoc
					Dept. Computer Science, Korea Univ.                        Intelligent Information System Lab.




                    User preference based
                    automated selection of web
                    service composition

                     Sohn Jong-Soo

                     Intelligent Information System lab.
                     Department of Computer Science
                     Korea University

                                                                                 1
Dept. Computer Science, Korea Univ.    Intelligent Information System Lab.




          Index
          1. Introduction
          2. Modeling aggregation information
          3. User preference modeling
          4. Automated plan selection
          5. Conclusion and outlook




                                                             2
Dept. Computer Science, Korea Univ.                                       Intelligent Information System Lab.




          1. Introduction
           End users compose web services because
                 ■ there may be no single web service
                         that directly offers the desired functionally
                 ■ a combination of web services may need less investment or
                   capabilities than a single service
           Need for machine support to perform calculations
                 ■ the number of web service combinations can be large
                 ■ the structure of web service combinations can be complex
                 ■ user may to do such calculations very often
           Focus is on the selection phase
                 ■ propose to model aggregation



                                                                                                3
Dept. Computer Science, Korea Univ.                        Intelligent Information System Lab.




          2. Modeling aggregation Information
           2.1 Description of Non-Functional Properties
                 ■ most common non-functional properties


           2.2 Description of web service combinations
                 ■ modeling four type of compositions


           2.3 Modeling aggregation information




                                                                                 4
Dept. Computer Science, Korea Univ.                                  Intelligent Information System Lab.




          2.1 Description of Non-Functional properties

           QoS
                 ■ define a minimal level of service of quality to provide
                 ■ Locative Availability
                 ■ Availability Rate
           Price
                 ■ monetary amount
                 ■ Amount
                 ■ Currency
           Payment method
                 ■ Duration
                 ■ Charging Style



                                                                                           5
Dept. Computer Science, Korea Univ.                           Intelligent Information System Lab.




          2.1 Description of Non-Functional properties

           Security
                 ■ Identification Methods
                 ■ Encryption Methods
           Trust
                 ■ trust levels with respect to web service
                 ■ Rating
           Privacy
                 ■ Storage Period
                 ■ Disclosure
                         yes or no




                                                                                    6
Dept. Computer Science, Korea Univ.                  Intelligent Information System Lab.




          2.2 Description of web service combinations

           Four types of web service compositions




                                                                           7
Dept. Computer Science, Korea Univ.       Intelligent Information System Lab.




          2.3 Modeling aggregation information

           Aggregation functions




                                                                8
Dept. Computer Science, Korea Univ.                   Intelligent Information System Lab.




          3. User preference modeling
           User preference can be modeled with
                 ■ fuzzy IF-THEN rules


           Modeling of fuzzy membership functions
           How can be calculated inside an appropriate
            description logic reasoner




                                                                            9
Dept. Computer Science, Korea Univ.         Intelligent Information System Lab.




          3.1 Modeling fuzzy membership functions
           Fig. 2 shows linguistic terms
           Defining the concept Point




           Set of membership function μ




                                                                 10
Dept. Computer Science, Korea Univ.                                 Intelligent Information System Lab.




          3.2 Modeling fuzzy rules
           A fuzzy IF-THEN rule consists
                 ■ IF part (antecedent)
                         combination of terms
                 ■ THEN part (consequent)
                         combined by using fuzzy
                                      conjunction
                                      disjunction
                                      negation

           Term representation
                 ■ the elementary building blocks of a fuzzy rule
                 ■ modeling Term




                                                                                         11
Dept. Computer Science, Korea Univ.    Intelligent Information System Lab.




          3.2 Modeling fuzzy rules
           Term expression TermExp




           Defining concept of Rule




                                                            12
Dept. Computer Science, Korea Univ.                Intelligent Information System Lab.




          3.3 Calculating the degree of fulfillment of a rule

           Concept defining




                                                                        13
Dept. Computer Science, Korea Univ.                                        Intelligent Information System Lab.




          3.4 Modeling user’s preference as fuzzy rule

           Modeling user preference by using IF-THEN rule
                 ■ IF part
                         membership functions of the various properties of an individual
                 ■ THEN part
                         one of the membership functions of a special concept called
                          Rank


           An example fuzzy IF-THEN rule




                                                                                                14
Dept. Computer Science, Korea Univ.                                          Intelligent Information System Lab.




          4. Automated plan selection
           4.1 FITA (first inference then aggregation)
                 ■ inference step
                         for given x and a given rule I means calculating

                         where F is some fuzzy set describing the membership function
                          for given situation


                 ■ in the aggregation step


                 ■ defining concept of FITA




                                                                                                  15
Dept. Computer Science, Korea Univ.                                Intelligent Information System Lab.




          4. Automated plan selection
           4.2 Defuzzification
                 ■ instance w can be calculated by the following formula :



                         using center of gravity method


                 ■ defining a concept defuzzification information, DefuzInfo


           4.3 Calculation of Ranking




                                                                                        16
Dept. Computer Science, Korea Univ.                                        Intelligent Information System Lab.




          5. Conclusion and outlook
           This paper presented
                 ■ a genetic approach for modeling aggregation information for
                   various web service attribute
                         as part of an ontology for describing composite web services
           Techniques in this paper
                 ■ presented web service combinations can be compared
                 ■ ranked according to the user preferences
           In the future
                 ■ to investigate the possibility of integrating multi attributive
                   negotiations




                                                                                                17
Dept. Computer Science, Korea Univ.                                                     Intelligent Information System Lab.




          5. Conclusion and outlook
           My impression
                 ■ 웹 서비스에 필요한 속성 및 설정을 표현
                 ■ 온톨로지로 표현하기 위한 작업
                         온톨로지로 표현하는 방법에 대한 기술이 없음
                 ■ 웹 서비스에 실제로 적용 시 표현된 속성과 설정이 어떻게 사
                   용될지에 대한 구체적인 예시 (mathematical)가 없음
                 ■ F-OWL (Fuzzy Web Ontology Language)
                         F-OWL에 대한 연구는 비교적 활발한 편이며 구체적이므로
                          웹 서비스에서 F-OWL을 활용하는 방안을 고려해보는 것이
                          적합하겠다고 판단됨
                         Fuzzy OWL: Uncertainty and the Semantic Web
                                      Giorgos Stoilos1, Giorgos Stamou1, Vassilis Tzouvaras1, Jeff Z. Pan2
                                      and Ian Horrocks2



                                                                                                             18
Dept. Computer Science, Korea Univ.               Intelligent Information System Lab.




                                      Thank you



                                                                       19

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:8/20/2012
language:
pages:19