Research Of RkNN In The Application Of Spatial Databases

With the rapid development of Internet and the Web, there is an abundance of information provided in various ways. It has become a hotspot how to find the information accurately and promptly from massive resources in the network. In fact, while users are suffering from searching what they need in such massive information, providers also want their information quickly received by those interested in it. An effective method is to release information only to users who might have interest in it.As the development of spatial information technology, application of geographic information systems has become wider and wider. As one of the most important branches of spatial science, certain RkNN is required in practice. RkNN is popular in intelligent navigation, modern communications, traffic control, weather forecasting and other areas.In practice, most data (over 85%) has spatial attributes, such as addresses, telephone numbers, Customer data distribution or distribution of assets, etc. Such attributes can be used to observe the trend and to grasp the opportunity. In short, swift and effective management of spatial data, attribute their space analysis. It is absolutely necessary for today's enterprises to analyze the attributes of spatial data and arrange the data effectively and promptly. What's more, such spatial attribute play a more important role than data itself, especially in fields of telecommunications, urban planning, public communications, petroleum geology, and so on.Spatial database has its own specificity. Firstly, the structure of data objects is complex. It is because that there are many other non-spatial attributes in the description of the object apart from the geometric information. As a result, spatial databases are usually very great. Secondly, data objects are often dynamic, with insertion, deletion and update staggered. Since it is difficult to find a unified standard for the sequence of spatial data, dynamic memory is quite difficult. Great importance has been attached to RkNN search in multidimensional spatial database, for example in GIS, CAD, multimedia data, and in genome research. However, the existing RkNN algorithms can generally get approximate results, or is applied only in certain specified circumstances. So in some extent there are deficiencies :Some can be applies only to the solution of two-dimensional spatial data, with the value of k restricted; Some support only static databases but dynamic updating, particularly moving objects database; While some algorithms don't lead to accurate but approximate results.Compared with the RkNN issues, kNN is easier to achieve, not only because it's studied earlier, but also because it has formed relative mature theory and algorithms. Under normal circumstances, there is no direct relationship between kNN and RkNN. That is r ∈ NN_s(q) does not mean r ∈ RNNS (q). But it can be proved that in certain conditions, RNN(q)(?) NN (q) and RkNN (q) (?) kNN (q).It means that we can put a RkNN solution into a kNN solution, or solve a RkNN problem through relatively mature kNN algorithm. Solving RkNN ultimately achieve the purpose. Based on this, this paper presents a RkNN algorithm using of the kNN in hand, which can support updating of the data in three-dimensional space and is suitable for any k value.