UML Tool Evaluation Requirements Ana Funes, Aristides Dasso, Carlos Salgado, Mario Peralta Software Engineering Group (SEG) http//www.sel.unsl.edu.ar Universidad Nacional de San Luis D5700HHW San Luis, Argentina {afunes, arisdas, csalgado, mperalta}@unsl.edu.ar Abstract. The Unified Modeling Language is widely used today in different stages of system development. A growing number of tools that support the UML notation are available. Evaluation of these tools is important so as to be able to choose among them the best suited to the job. Designing and developing a metric to conduct such evaluation is the first step in the assessment process. To be able to construct such a metric we have studied different desirable char- acteristics, facilities, behaviour and performance that such tools should meet. All those towards constructing a model that can yield a quantitative indicator of requirement satisfaction when it is applied to the tools under evaluation. We discuss the different features of the proposed Requirement Tree and the reason- ing behind the choices made as well as an example of an Elementary Criteria. The Requirement Tree and the Elementary Criteria both are elements in the first stages of the LSP method –the evaluation method chosen for the task at hand. 1 Introduction The Unified Modelling Language (UML) [OMG03], [BRJ98], [RJB98] has lately be- come a defacto standard for software developers, whether for documenting an exist- ing system, when doing reverse engineering on legacy systems or for development purposes from the early stages of development up to and including coding. There are several tools that help the use of UML in any of its different applications. Users of those tools must evaluate and compare different versions of the tools they in- tend to use or are using to assess the possibility of changing or acquiring one. There are several ways to perform this evaluation from the simple rule-of-thumb to numeric or quantitative methods. Evaluating families of software tools such as database management systems, pro- gramming languages, web browsers, operating systems, etc., or any other kind of soft- ware tools, is done to choose one particular system among several possibilities or sim- ply to assess several software systems. Although this activity can have a great economic impact it is not always carry out with the care it should. There are several methods to do this evaluation ranging from the most informal to the more careful and formal, from the simpler form based on the personal opinion of evaluators, to the one that using the opinion of evaluators or users