Research On Event Detection Method And Services Load Balancing Based On Graph Convolutional Network

Event detection is dedicated to judging whether natural text contains events and identifying their types.As a subtask of event extraction,it plays an increasingly important role in data engineering and other directions.Most of the current research on event detection focuses on the improvement of event detection effects and performance.The methods involved in these studies include traditional pattern matching methods,machine learning algorithms and popular deep learning models.These methods have a wide variety,different ideas and different application conditions.The above characteristics have brought great difficulties and challenges to the reproduction and landing practice of each modelConsidering that the distributed,easy to deploy,easy to reuse,easy to iterate and easy to expand characteristics of microservices correspond to the above-mentioned various models,this paper proposes to microservice each event detection model and other commonly used strategies and datasets.An event detection system based on microservice architecture is designed,and a multi-objective microservice load balancing strategy is proposed,which solves the problem of service response efficiency of the event detection model by means of microservices.The main works of this paper are as follows:(1)An event detection method based on attention graph convolutional network is proposed,which improves the effect of the event detection model.The GCNs models that has achieved excellent results in the current event detection problem is improved.First,the edge label information of the syntactic dependency tree is introduced into the model,and then a scoring strategy based on attention weight is designed as the edge update weight.Finally,the robustness experiments and analysis of the GCNs event detection models are carried out.The experimental results show that the method improves the effect of event detection to 78.4%,ranking first among GCNs event detection methods and second among all known event detection methods.It also has better robustness than other GCNs event detection models.(2)This paper proposes to microservice each event detection model,and designs an event detection system based on microservice architecture.Design each event detection method or general preprocessing strategy as a microservice,and provide an open interface for users or other services to call.At the same time,a multi-objective microservice load balancing strategy is proposed,and a microservice user requests allocation model is established with the total request processing time,load balancing rate and total communication transmission distance as multiple objects.The allocation strategy of concurrent user requests among multiple microservice instances deployed in different resource centers is studied,and the problem is solved using a multi-objective evolutionary algorithm based on a custom initial solution generation strategy,crossover operator and mutation operator.The experimental results show that the above strategy realizes the load balancing of the event detection microservice system and improves the service response efficiency of the event detection model.