The Matching Optimization Of Geo-textual Object Stream With Subscription Stream And The Management Of Subscription Stream

After entering the 21 st century, mobile Internet and global positioning technology have been improved quickly, a growing number of text messages have been attached with geo-tags, a text message attached with geo-tag is called a geo-textual object. Sometimes a user needs to find the specific geo-textual object from large mounts of constantly producing objects. Regular queries on geo-textual object database system are executed for the latest information, which causes inconvenience on the user and also increases the capacity of the system.Therefore, a new geo-textual object distribution subscription system is necessary, a subscription corresponds to a user's request, its main function is to manage the upcoming geo-textual object stream and the upcoming subscription stream. The main tasks include matching the geo-textual object stream with the subscription stream and managing subscription stream.To match the geo-textual object stream with the subscription stream, a hybrid indexing tree is used to organize the data objects, including the multi tree space index, predicate index and inverted file. Multi-tree space index is built for spatial area management, predicate index and inverted file are used to manage the predicates.To manage the subscription stream, we use a kind of 3-level subscription buffer model, ie. Memory, solid state disk and mechanical hard disk. And the scheduling policies on the caches of Memory-Solid State Disk and Solid State Disk-Mechanical Hard Disk are also appied. The 3-level subscription buffer model is used to solve the problem of the storage of subscriptions, and two scheduling policies above are used to handle the scheduling for subscriptions on different storage medium.This paper elaborates the core technologies of our system in detail, then conducts multi-group contrast experiments under the environment of real data. Finally, the experimental results verify the feasibility of our system, and further analyze the effect of key technical parameters on the performance of the system.