A Study of a Framework of Integration of Heterogeneous Spatio-Temporal Data
3.2 Framework of data integration
We will show framework of data integration based on the preparation in previous chapter. The framework consist of three steps:
- Retrieval and acquisition of the observational data
- extraction of observational data on the object
- Estimation of the state/dynamics of the object from observational data
3.2.1 Retrieval and acquisition of observational data
This is a process of retrieval and acquisition of observational data, which may relate to an object requested by a user, form existing various data set by referring to its quality and product specification by way of each metadata. We can raise some examples of criteria ofr4 retrieval and acquisition .
-
Distance of hierarchy along hierarchy of object definition ( for all objects
- Distance of concept ( for all objects )
- Distance in space and time ( especially for continuous distribution object ).
3.2.2Extraction of observational data on the object
We classify observational data,. Collected as data related to the object into two types: one is data obtained room observing the object. The other is not from the object. For example, map data is observational data already classified. However, even fro the map data this process is needed, because it might happen that object definition used in ht classification differs from that used by users. The classification method is basically a test of hypo9thesis whether the observational data is data on the object or not. The measure , which shows the "degree" that observational data is observed from object, is a probability of observation executed and a error generated by observation. These processes of classification enable us to extract observational data of object.
3.2.3 Estimation of the state/dynamics of the object from the observational data
We estimate location, shape , vector of motion, and an attribute of object using classified observational data in 3.2.2 . first we set a function of object that consists of and attribute of object and coordinates of point s representing a shape of object : The function is [ (xi (t), yi(t) (I=1,2,…..,N)+Attribute (t) ] . in the next stage we assume each data have error and reliability by referring to its quality. For example features such as line can have probabilistic distribution . And moreover, estimation reliability can be improved by considering the constraints, which come form the shape and nature of object itself , in addition to the observational data. E can obtain constraints referring to the feature catalogue prepared in advance. Under these preparation we can estimate object from some data by statistical method.
3.3 Information needed for data integration
As considered before, we refer to many kinds of information in each step of data integration. They are classified below:
- Information about product specification of data
-
Information about data (metadata)
-
Information about feature catalogue
3.3.1 Information about product specification of data
Product specification of data may include definition of object describing criteria which data should satisfy, required accuracy of location, shape , an attribute, name of feature cataloguer to be referred and source of information to be referred. Product specification is accessed by way of metadata.
3.3.2 Information about data (metadata )
Metadata should contain the information on quality of data such s accuracy of location, shape and classification
3.3.3 Information about feature catalogue
Feature catalogue, which is accessed by way of metadata, consists of feature catalogue peculiar to product specification and common feature catalogue. It is expected that these be arranged as a part of NSDI.
4. Future Perspectives
We propose a framework of spatio-temporal data integration, therefore methods of each step should be developed. First we plan to focus on the case of immobile .
Reference:
-
Kadowaki, T., and shiibasaki, R., 1992, Measurement and evaluation of location error of line data on digital map. Proceeding of the symposium of the Japan Society of Photogrammetry and Remote Sensing, 1992, pp.71-76.
