Study The Application And Research Of Hidden Markov Model In Chinese Geo-Entity

With the blooming or the Internet, especially the mobile internet, the speed of information distribution increase dramatically. The distance between people is being shortened. The real work is being virtualized gradually. People use Internet to make friends, do shopping, and search all kinds of delicious foods and good deals. By the combination of Internet and real work, the life experience has improved a lot. Meanwhile, in order to cover this increasing need, may applications based on GIS are published.All kinds of GIS applications and location based service all needs to process a large amount of data. During those data, the entity data in the map (we also call them POI-Point of Interest), are playing the key role of the process. The quality of those GIS application and location based service is decided by how those data can be processed the data properly. Before those GIS entity are used, there are usually a series of processes, i.e. normalization, classification, and duplication delete process. Those processes affect the usage and search of data by user, and affect user experience. During the series of process, doing the classification properly and correctly is key and hard point. Many problems are need to research, especially for Chinese GIS data. Same entity may have different names. The abbreviation for Chinese Entity name is also different. Meanwhile the speech-check for Chinese POI is also special problem.This paper does a deep research on classification of Chinese POI. Firstly, this paper researches a kind of Chinese word segment, which is apt for Chinese POI classification problem. This paper also compares various classifiers, and then applies Native Bayes classifier to the problem. Then based on the words relation research, Hidden Markov Model is used to solve the Chinese POI classification problem. Results from those two classifiers are compared. The abbreviation rule for POI name is also studied in this paper.