| Since ontology was widely applied in the field of AI,this has promoted the development of intelligent technology. To describe knowledge within the field through use of ontology, however, changes as time goes by. In the business environment of dynamic development, the amendment of application program, the change of business strategy and customer demand will bring about a result that original ontology is not adequate for application requirement, and thus lead to the research on problem of ontology evolution. Ontology evolution is to manage changing ontology, while its purpose is to change correspondingly for other ontology data which is relevant to this ontology after ontology happen to change, keep semantic integrity for all ontology after varying and data accuracy, and finish use requirement of application program.PI(π) calculus as one kind of process algebra is thought by Robin Milner and other people. Robin Milner invents PI calculus on the base of CSS. PI calculus with a focus on concurrency theory is proposed by mobile communication based on inter-process. As a tool, PI calculs is able to establish a new channel, so it is called mobile process algebra. It is used to description and analysis system interactions between multiple components, in particular dynamic system.Ontology evolution is a course that ontology changes to adapt environment by it self for achieving the request of apply. It has the characteristics of dynamic change, while PI calculus has ability to describe a system that is often changed. Based on this point, author considers using PI calculus formalization of ontology evolution, using PI calculus to describe the dynamic process of ontology evolution.First of all, the article summarizes the methods of existing ontology evolution. especially introduce detail of user-driven(6-phase)evolution method. Six-phase is change capturing phase, change representation phase, change semantic phase, change propagation phase, change implementation, change validation phase. Their functions, methods and other related knowledge have been explained. Especially evolution strategy is given the necessary elaboration in the change semantic phase. This thing is in favor of bringing forward based on"semantic-driven evolution strategy"next chapter.Secondly, the article summarizes and compares several advanced evolution strategies: structure-driven, process-driven, instance-driven, frequency-driven evolution strategies. They have some shortcoming thus author presents a new evolution strategy: semantic- driven strategy. By using three concepts including"semantic distance","semantic weight" and"resistance"which can describe semantic information about ontology entities, the article give a strategy called semantic-driven strategy and algorithm process correspond with this strategy. This strategy can be applied in according with the semantics of ontology to build a flexible evolution strategy. The ontology will change into another new ontology that is consistent and according with the need of application. This strategy is also cooperate with other advance strategy. The result caused by merge is used for ontology in special domain changing. Get a new ontology that is working out special requirements. Use a ontology model to exhibit how the"semantic-driven ontology strategy"works step by step at the end in this chapter. Through a series of processer, we can know that ontology entity based query judge whether it is according with the domain knowledge, and then it is able to change itself dynamic before it chooses different strategy elements. The case prove semantic-driven strategy is flexible because it is base on the semantic of the ontology entity in the domain. It would be in collaboration with other advance strategy is also testified. Again, on the basis of PI calculus, EPI calculus is presented. It extends PI calculus.There is a new entity-"term"in EPI calculus. Article give the syntax and operational semantics meaning of the two entities-process and term. Also three basic reduction rule are work out. Using EPI calculus models service consistent system in the Web service registry center. After that, make EPI calculus awareness and understanding more intuitive and show that EPI calculus has descriptive modeling capabilities.Finally, EPI calculus formalized representation of ontology evolution process. Give the constraints to build"EPI instance model", the mapping rules between ontology entities and EPI calculus entities. Through the constraints and the mapping rules, the EPI instance model change into"EPI formal model". Since then, use the EPI calculus describes process how to create, delete and modify the channel. The operation process and ontology process are modeled by calculus. use of a simple ontology model, EPI calculus formal describe the system of EPI formal model about this ontology model. Using reduction rules exhibit the process of the system how to change step by step. Apply the reduction rules to change the new EPI formal model into ontology model at the end. This ontology model is consistent and answered for ontology evolution by strategy guided. All this things prove that EPI calculus has ability of formal describing ontology evolution. It has the function of model evolution.There are some shortages in this article. For example, it has no actual instance about ontology evolution to study six-phase function; For the new strategy"semantic-driven ontology strategy", it has not done in the actual application, just only given the level of algorithms and ideas in theory; though EPI calculus as a tool can formal system of ontology evolution how to dynamic change, the characters about EPI calculus, such as structural congruence, bisimulation and other theory are not researched. The article only study in the domain of application with EPI calculus.The future work will be aimed at two aspects of ontology evolution and calculus. Ontology evolution will focus on improving its method, propounding appropriate new algorithm in view of functions which should be finished for each stage, and confirming the correctness and feasibility of improvement though specific data by comparing with current method.Calculus will focus on study in theory about the new calculus which is get by extend existing calculus. After research calculus invented by other people, like PI, Ambient, SPI calculus, thinking about the characters of those calculus and actual question The part of calculus emphasis is put on theory research that it obtains new calculus after expanding current calculus. It is to combine particularity of the field as described, with deeply study of past calculus such as PI, Ambient, and SPI calculus, etc, and thus compound basic definition of our own computational language, new calculus entity, grammatical representation of entity and semantic meaning. Meanwhile on this basis, the theory of structural congruence, bisimulation reduction rules etc on new calculus should be deeply researched. This is to prove what properties of past calculus have been kept and what has been improved already.The ultimate research goal is put ontology evolution with calculus improved on other calculus by myself together. Use calculus to describe the whole process of ontology evolution and other knowledge that is according with the characters of the calculus. Besides that, consider to use the own characters of the calculus to judge whether ontology gained by evolution are similar. Such that Judging whether ontology are the same on the level of the semantic by Structural congruence, bisimulation, etc. |
PDF Full Text Request