Abstract
In this research, we propose a dynamic decision making using social context based on ontology. Dynamic adaptation is adopted for the high qualified decision making, which is defined as creation of proper information using contexts depending on decision maker's state of affairs in ubiquitous computing environment. Thereby, the context for the dynamic adaptation is classified as a static, dynamic and social context. Static context contains personal explicit information like demographic data. Dynamic context like weather or traffic information is provided by external information service provider. Finally, social context implies much more implicit knowledge such as social relationship than the other two-type context, but it is not easy to extract any implied tacit knowledge as well as generalized rules from the information. So, it was not easy for the social context to apply into dynamic adaptation. In this light, we tried the social context into the dynamic adaptation to generate context-appropriate personalized information. It is necessary to build modeling methodology to adopt dynamic adaptation using the context. The proposed context modeling used ontology and cases which are best to represent tacit and unstructured knowledge such as social context. Case-based reasoning and constraint satisfaction problem is applied into the dynamic decision making system for the dynamic adaption. Case-based reasoning is used case to represent the context including social, dynamic and static and to extract personalized knowledge from the personalized case-base. Constraint satisfaction problem is used when the selected case through the case-based reasoning needs dynamic adaptation, since it is usual to adapt the selected case because context can be changed timely according to environment status. The case-base reasoning adopts problem context for effective representation of static, dynamic and social context, which use a case structure with index and solution and problem ontology of decision maker. The case is stored in case-base as a repository of a decision maker's personal experience and knowledge. The constraint satisfaction problem use solution ontology which is extracted from collective intelligence which is generalized from solutions of decision makers. The solution ontology is retrieved to find proper solution depending on the decision maker's context when it is necessary. At the same time, dynamic adaptation is applied to adapt the selected case using solution ontology. The decision making process is comprised of following steps. First, whenever the system aware new context, the system converses the context into problem context ontology with case structure. Any context is defined by a case with a formal knowledge representation structure. Thereby, social context as implicit knowledge is also represented a formal form like a case. In addition, for the context modeling, ontology is also adopted. Second, we select a proper case as a decision making solution from decision maker's personal case-base. We convince that the selected case should be the best case depending on context related to decision maker's current status as well as decision maker's requirements. However, it is possible to change the environment and context around the decision maker and it is necessary to adapt the selected case. Third, if the selected case is not available or the decision maker doesn't satisfy according to the newly arrived context, then constraint satisfaction problem and solution ontology is applied to derive new solution for the decision maker. The constraint satisfaction problem uses to the previously selected case to adopt and solution ontology. The verification of the proposed methodology is processed by searching a meeting place according to the decision maker's requirements and context, the extracted solution shows the satisfaction depending on meeting purpose.
본 연구는 사용자의 정적, 외부환경과 연관된 동적 상황정보와 사회적 관계와 연관된 개인적 상황정보들을 의사결정 요소로서 고려한 의사결정의 동적 변환(Dynamic Adaptation)을 제안한다. 즉, 의사결정자의 정적, 외재적 정보보다 과거의 경험, 주관적 선호도 및 사회적 관계와 연관된 상황정보(Social Context)를 의사결정에 동적으로 반영하고 동시에 의사결정 해의 사용시점에서의 가용성에 따라 유용 가능한 대안을 추출하는 방법론을 제안하고자 한다. 이를 위해, 정적, 외재적 및 사회적 상황정보를 이용하여 의사결정 추론한다. 추론은 의사결정자의 과거 경험에 기반한 사례기반 추론과 해당 의사결정 결과가 가용하지 않을 경우 수정을 위한 제약식 만족추론으로 이루어진다. 이를 위해 개인적 경험 등의 정보에 기반한 '문제상황 온톨로지'(Problem Context Ontology)와 집단의 경험적 지식에 기반한 '솔루션 온톨로지'(Solution Ontology)를 구축하였다. 의사결정단계는 상황정보 인식 및 문제상황 온톨로지에 매핑하는 단계, 경험적 사례로부터 문제상황에 가장 적합한 사례를 선택하는 단계, 생성된 솔루션이 가용하지 않을 경우 솔루션 온톨로지와 제약식 만족추론을 통해 새로운 대안을 생성하는 단계로 이루어진다. 본 방법론을 모임에 적합한 식당을 제안하는 예제를 적용함으로써 타당성을 검증하였다. 또한 실험을 통해 사회적 상황정보를 고려하여 생성된 의사결정대안이 그렇지 않은 경우보다 의사결정자의 만족도를 향상시켰으며, 생성된 의사결정대안이 가용하지 않은 경우 제약조건식과 솔루션 온톨로지를 이용해 생성한 대안이 유의미함을 검증하였다.
