Broadcast Data Based On The Active Mechanisms Of Rapid Filtration Method Of Research

Data broadcast is a new field of information service in recent years, it is a new dissemination tool and integrate the internet and broadcast television networks.Its source can be from every web site, its channel is digital television,and it push all the content of the web site to customer's computer.It has both the advantages of the two media above,just as has adequate signal source and can spread abroad. With the develepment of the data broadcast, there must have a system can help customer to select what his want from the giant information.We put forward this project to meet this demand.The goal of our porject is researching and implementing a system which can receive and handle the broadcasting data.At first we simulate the data broadcast server and client based on the theory of multicast and data broadcast. Then we expand the activity rule and use it to help us manage the receiving system.At last we design a quick filter which can filter the useless imformation at the very start.Active Database can response for the changing of state in or out of system without the control of customer. So we expand it to help manage our system .We design the event,action and rule according to the demand of data receiving system,then use two level of priority to handle the rule conflicting.We improve the flexibility of our system through encapsulating event detector into DLL.There are two main methods of text filtering.One is character string matching, this method is fastest but its precision is lower.The other is revised classfication model,which has high precision but slow speed. In this paper ,we use two level to filter document.At first level ,we use Aho-Corasick to filter data packet ,then at second level we use concept hierarchy and CHI to filter the text.At the end of the paper,we put up a experiment . Via the experiment ,we conclud that,our filter method have high precision and fast speed.