Vector Data Contrasting Technology Research Based On Probabilistic Theory And Compound Criteria

In the real world, geographic data of the same area or city is collected by different people at different time. Then, the respective map databases are set up. The most urgent problem in the current GIS is how to realize the fusion and integration of the data, which comes from different areas at different degrees. At last, new dataset is generated for the subsequent research. Here, digital map conflation technology is always adopted.Digital map conflation is widely used in the field of geographic information data fusion, integration and incremental analysis. Also, it is a difficult point in GIS engineering construction. There are often two processes in digital map conflation—entity matching and merging data, which means geometry, attribute and graphic information of the entities will be contrasted firstly and then different data from different places will be fused and integrated. Based on entity matching, relevant attribute and graphic information of two datasets will be obtained through this technology, and the last step is the two datasets fusion. In this process, one of the maps is called "source map" or "the reference", and the other one is called "destination map" or "the adjustment". Both processes are very important and cannot be dispensed. Moreover, entity matching can be seen as the first step or basis of digital map conflation, and the subsequent merging is the finally source of digital map conflation results.Identical entities are in two maps or two databases, which are a kind of spatial entities reflecting the same feature or features in the real world. Through analyzing spatial entities' difference and similarity, identical entities recognition or matching is used to distinguish the same entity in the datasets from different places.Traditional map entity matching methods mainly discuss about the spatial information, that is, contrasting the similarity and difference of entities'spatial position, shape and direction and other information. However, in the database of GIS, vector data's attribute, spatial and graphic information should be comprehensively contrasted and analyzed so that the best contrasting results will be obtained to realize the next spatial incremental updating work.Based on the domestic and foreign scholars'research, this paper analyses the advantages and disadvantages of current map matching methods combined with national "863" program—"distributed heterogeneous-geography seamless integration supporting incremental updating research and software development". Furthermore, based on the front GIS development, it discusses new theories, methods and technologies of current data contrasting, and proposes vector data contrasting technology research based on probabilistic theory and compound criteria, which comprehensively considers vector data's attribute, spatial and graphic information and many kinds of criteria are involved to efficiently solve 1-many or many-many matching situation. The results represent that this method has good recall and precision, and can be used in vector data's incremental updating.According to the contrasting technology's characteristics, this paper will start the research from the following aspects:Chapter 1 introduces domestic and foreign entity matching methods' current situation in details. It analyses the advantages of current methods and concludes current existing disadvantages which need to be solved urgently. Besides, the necessity and feasibility of discussing and studying new matching methods are proposed.Chapter 2 systematically explores digital map conflation, including the basic concepts, studying content, category, tech-process and the relationship with other technologies.Chapter 3 analyses vector data in digital map conflation, including attribute information and spatial information. According to the characteristics of entities' attribute information, attribute difference is classified in details to design entities' attribute structure rules and attribute information contrasting methods. Besides, choosing the criteria and computing methods of spatial information is introduced in details. Also, it explores the method of how to choose the threshold for the same entity with different criteria weights.Chapter 4 concludes all non-probabilistic matching methods in entity matching, and analyses traditional matching method based on probabilistic theory and vector data contrasting technology research based on probabilistic theory and compound criteria. It introduces candidate sets' confirming, weights' choosing and probability computing of entities or datasets. According to entities' spatial characteristic classification, the new method refines criteria computing, fuses attribute contrasting and improves threshold choosing.Chapter 5 introduces the design and realization of vector data contrasting in MapGIS K9. First, it introduces design concepts and the whole framework of vector data contrasting. And then, it discusses the main interfaces which need to be added in geographic database to realize the vector data contrasting module. At last, the screenshots and results' analysis report are provided.