Data Mining Research Of Plotted Data Based On Spatial And Temporal Association Rules

In recent years, the Internet has come into a mobile era, people now used to browsing and sharing information through applications installed on their mobile devices. The geo-referenced marks made by people are called plotted data. Users have generated a huge amount of plotted data while they are using mobile applications which are location based services. The ever growing volume of plotted data has shed light upon many research fields, bringing challenges and opportunities to researchers.In contrast to other data, the plotted data has unique characteristics. The geographic position and temporal information are naturally contained. Owing to the development and promotion of science and technology, a huge amount of plotted data is growing rapidly and the structure of plotted data is more multiple and complicated, the quality of it is highly heterogeneous. But all of these characteristics show that the value of plotted data is great and useful. Using data mining approaches can help us find the meaningful knowledge to provide better service.First of all, the middleware of getting plotted data is achieved through packaged HTTP based on the researches Web data acquisition methods and the plotted data is preserved by utilizing Post GIS spatial database on the basis of plotted data characters. Then the plotted data is preliminary screening by using attributes of itself or related data, such as POIs, and in order to make the plotted data cleaner, converting the unstructured text information to the text feature vector by using technique of Chinese lexical analysis. Realizing the subject oriented extraction of plotted data through relation database keywords screening method. At last, hotspots and association rules mining of plotted data are presented based on the K-means algorithm, kernel density estimation method and the Apriori algorithm. Using the methods to plotted data can find the hotspots regional distribution in space and the co-occurrence associated patterns of spatial-temporal event.Based on the methods proposed above, taking the geo-located micro-blog data of Sina as an example. The distribution and situation of tourism hotspots in May Day and National Day are analyzed by using hotspots mining methods. The possible rules between tourist origins and destinations, tour time and spatial relation are exploited by utilizing spatial-temporal association rules.The results show that the temporal and spatial attributes of plotted data can be applied in data mining research. The minds and the methods formed in this research can be used to find the hotspots and association rules of plotted data. The statuses and trends of holiday tourism hotspots can be effectively reflected by using hotspots mining methods. The relationships of destinations, passenger sources, travel time and spatial relations can be found by using spatial and temporal association rule mining. The analysis results can provide information to help making travel decisions.