Thursday, February 25, 2010

On spatial relations between events and entities

During the final pass of revisions for the EPIA book, we have re-visited the issue of spatial context, one of the options of spatial context is to group or partition events according to the relationship between event and a certain entity. However, location is not necessarily a point, there are three main type of location: point, line (or poly-line) and area (polygon). This is true for entity, but it is also true for an event, if the event is located using GPS, then a point coordinate is obtained, but if an event is located using the signal sent from a cellular phone, then the location of the event is a cell, i.e. it occurs within an area. We can also say that a vehicle is somewhere on road M1, which is a poly-line. Given these three data types, we can define several relations among the location of the event, and the location of the entity, as shown in the illustration above, such as: contains --- the entity contains the event, disjoint etc... The illustration above shows the various cases between events of various location types and entities of various location types. Some examples of use cases for the various relations will follow in one of the next postings, meanwhile -- you can think about applications yourself

1 comment:

Johannes Echterhoff said...


Nice article, thanks. You are probably aware that there are more types of spatial relationships. Also of interest would be the Dimensionally Extended 9 Intersection Model (DE-9IM.

More formal definitions of such operators can be found at the Open Geospatial Consortium (OGC) and ISO TC 211.

We have done some work on Spatial Event Processing at OGC, primarily in the domain of Sensor Web. However, I am very interested and also looking into Spatial Event Processing in general. There are some interesting applications where the combination of Event Processing and Spatial Data Infrastructures would be benefitial - Sensor Web being one of them.