In my presentation I will talk about symbolic images and direction relations. A symbolic image is an array of symbols, where each symbol is a representative of an object. The position of object symbols in the array indicates a set of spatial relations among them. We will then see how we can query a database of symbolic images by queries that are themselves symbolic images. Examples of queries with union, intersection and join will be shown and a simple geographic application of this system will be mentioned.

Questions people ask other people are rarely precise and even more rarely precisely answered. Unfortunately, queries written for computers must be very precise and frequently their results contain either too much information or no information at all. As computers hold an increasing amount of information and people increasingly use computers, reasonable ways of mediating information interchange are required. One method of solving this problem involves using the idea of preferences. Users define the information they would prefer to retrieve and are presented with information providing the closest match. Preference SQL is a method of using preferences in relational databases. Queries can contain predicates such as 'around' instead of '=' and provide results from the database 'matching' the query - perhaps exactly or perhaps loosely. We will discuss Preference SQL as well as other ideas from the preference literature.

In dynamic environments data sources may change not only their data but also their schemas and semantics, which causes serious problems in the systems using these data sources. The problem becomes more critical, with XML, becoming more popular as data storage stan- dard and mean for data exchange between various systems. While there are many algorithms that have been invented to deal with a problem, the issue of performance still remains a major concern in such kind of algorithms. In this proposal new idea is proposed to deal with the problem using summary of the database schema.

The talk will first motivate the need for federated systems. Two federated systems will be presented. The first system is using the idea of Garlic into IBM's DB2 Universal Database so as to enable DB2 to access data from diverse non-relational data sources. A novel method to calculate the cost function based on system availability, process and network latency at remote sources, and the load at the system's nodes will be presented. The second system, MOCHA (Middleware Based On a Code Shipping Architecture), will be presented briefly.