Study On Ontology Knowledge Construction Method Of Service-oriented Task Under Private Cloud Framework

With the rapid development of robot technology,service robots are playing a more and more important role in family life.However,the complex and changeable information in the family environment makes the service robot lack of a unified,normative and learning ability knowledge system.Therefore,this paper studies the ontology knowledge construction method of service-oriented task under the framework of private cloud.Ontology library based on ontology can efficiently organize and classify the messy knowledge and realize the standardized application of knowledge.Based on object detection and natural language processing(NLP),an ontology knowledge automatic addition mechanism is proposed to make ontology libraries have the learning ability.In order to realize the information sharing among multiple robots,a special private cloud server is built in this paper.The relationship between different ontology libraries is used to realize the integrated representation of ontology knowledge base.The combination of the three solves the problem that the service robot is difficult to use the information of home environment directly.This paper mainly completes the following work:First,the object ontology library is constructed to normalize the arrangement and classification of the object information,which solves the problem that the family environment information is messy and difficult for the robot to use.It provides a large priori knowledge base for the execution of robot service tasks.In order to solve the problem of information exchange between robot and ontology library,an ontology library interface based on Python language is developed.The main function of the interface is to parse,query,update and write ontology library.In consideration of the actual situation used by users,the UI interface of simple operation is specially designed and the interface function is encapsulated into visualization software to realize the knowledge operation of the ontology library.Second,an automatic knowledge addition mechanism of ontology library is proposed to solve the problem of insufficient learning ability of ontology library.First of all,the object detection model is used to obtain the class name of the unknown item,and the knowledge demand at the class level or the instance level is determined according to the service request.The instance level demand also needs to identify the main color of the detected image to determine the color of the item.The relevant text information is crawled by web crawler,the text information is preprocessed and the effective information is extracted by text classification model.The effect of text classification model will directly affect the accuracy and comprehensiveness of knowledge selection.Therefore,this paper adopts LSTM neural network as the classification model and specially designs experiments to compare with the traditional classification model.Based on relevant theoretical knowledge,reasonable evaluation indexes are designed by ourselves.Experiments were carried out to obtain the scores of different classification models.The experimental results proved that the effect of the text classification model was effectively improved by using the above network.Finally,the experimental verification of the whole adding process was carried out,and the knowledge acquired in the network was successfully added to the ontology library.Finally,a private cloud server is designed and built,a variety of data interaction channels between the client and the cloud are established,and a real-time knowledge transmission mechanism based on SSH protocol is proposed to realize the timely transmission of knowledge from the client to the cloud.The integration and hierarchical representation of ontology knowledge under the framework of private cloud is realized by making use of the interrelation between different ontology libraries and combining with the automatic knowledge addition mechanism of ontology libraries mentioned above.At the same time,a special simulation family environment and service robot model are built,and a unified and standardized knowledge system is formed by combining the interface program of the ontology library.Under the support of the system,experiments are carried out and the robot is assisted to execute and complete the specific service tasks.