CRITICAL ANALYSIS OF DESIGNING A GRAPHICAL QUERY LANGUAGE

During the last two decades, information has been stored in a plethora of different databases, locations and data models. In many cases, companies and organizations use data, that is stored in a variety of data models and remote database servers, which can be correlated. A challenge would be to define queries joining data from all those different sources. Thus, data can be found in Relational, Object-Relational, Object-Oriented, XML databases or even in text files. For each model there is a corresponding query language, i.e. SQL, SQL3, OQL and XQuery. Moreover, it is very tricky to join data from different sources and we have not found in the literature a query language that has been designed to use data in the same query from Relational, Object-Relational, Object-Oriented and XML databases. On the other hand, the syntax of all the above mentioned query languages, is complicated and demanding especially for novice users. The majority of computer users need only to learn how to complete simple work tasks, whereas the problems they have to solve are usually expressed in noncomputing terms. For this reason users prefer to use graphical query languages instead of a query language that has a textual syntax. In this paper, we present our conclusions in designing graphical query languages (GQLs). We have already designed, developed and evaluated, by organizing a controlled experiment, two GQLs, namely GOQL for Object-Oriented Database System and KINISIS for XML databases. The most important lesson that we learned is that it is very important the design of a graphical query language to be “free” of the underlying query language. Finally, we introduce a solution to the problem of joining data from different sources which stored in different database models, based on a graphical query language.