|
Online network analysis from heterogeneous datasets - Case study in London train network
As discussed previously, the names of features coded in the rules are the representative name of feature types which can be defined in the same common group. In order to link back to the database/dataset name used in an application, the document of semantic links is required. The sample of coding semantic links follows. 
The relationship of documents which are required for each project can be shown in figure 3. 
Figure 3. The software system architecture As shown in figure 3, project document contains three separate XML documents that are interrelated. The project file contains a list of datasets required for the application. The family file supplies the rules of feature connectivity. The semantic schema provides the name mapping to the names of datasets used in the application. The link between project and family and family and semantic can be coded respectively as follows. 
4.2 Intelligent integration engine The intelligent engine to integrate data from multiple heterogeneous data sources via the web was developed using Java servlet technology. It allows users to send a request of which datasets they want via the web. When it receives a user request, it opens the project document and starts to open the web connection to the GIS data server defined in the project and return that data back to the system. When it has retrieved all the data required then it combines those data together. It also converts the rules of feature connectivity defined in the project document into a form which is understandable to the application. After getting all data required, it delivers these objects back to the application. 4.3 Network topology builder Network topology needs to be built to perform network analysis. It has to be built based on the rules of feature connectivity. It starts by cleaning geometries by intersecting and snapping the features and then builds the topology which contains a set of links and nodes for network family. With a set of links and nodes generated, this allows the application to perform network analysis. This tool was developed using Java technology. 4.4 Network analysis tool The network analysis tool was developed which enables a user to find the shortest path between two points. The shortest path function was developed by implementing Dijkstra's (Dijkstra, 1959) algorithm. To determine the shortest path, the weight of the path is required. In general applications, the weight is usually the distance of the link between two nodes. However the weight can be any number attributes used to find the shortest path e.g. travelling time. |