Integrating graphical information system models with visualization techniques

Abstract

Elektronisessa muodossa olevan tiedon määrä kasvaa jatkuvasti. On yhä vaikeampaa ymmärtää yksittäisen tiedon suhdetta kokonaisuuteen ja muihin tietoihin. Sama ongelma esiintyy myös tietojärjestelmäkuvauksissa. Tietojärjestelmäkuvauksia tarvitaan kommunikoinnin helpottamiseksi ja monimutkaisuuden vähentämiseksi. Kuvaukset ilmentävät tulevan tai nykyjärjestelmän rakennetta ja käyttäytymistä eri tasoilla. Tasoja voidaan tarkastella muun muassa yleisestä yksityiskohtiin tai vaihejaon mukaan analyysistä suunnitteluun ja toteutukseen. Huotari selvitti väitöstutkimuksessaan voidaanko visualisointitekniikoilla auttaa tietojärjestelmäkuvauksia arvioivia henkilöitä löytämään kuvauksista virheitä, puutteita tai ristiriitaisuuksia. Huotari perehtyi muun muassa tietojärjestelmien kehittämisongelmiin, informaation visualisointiin ja kognitiivisiin tekijöihin.

Graphical information system (IS) models comprise of several diagram types representing both static structure and dynamic behavior at many levels of detail. Critical design information is distributed across a collection of diagrams, which share some common elements such as objects and their properties. The examination of multiple diagrams slows search performance and could result in reading errors that later cause omissions and inconsistencies. We need better methods and tools, which help understanding the “whole” model and how different parts relate to each other. Especially, during the reviewing process, reviewers need representations that help exploring the model and enable seeing inter-relationships between diagrams. In this thesis, we examine visualization and cognitive aspects of representing design information. We start this by synthesizing the main contributions and research methods in the HCI field. Then we propose a framework for evaluating visualization tools. As a third step, we introduce initial findings and suggestions for using advanced visualization techniques in computer-aided software engineering (CASE) environments. Our first research prototype was used in a laboratory experiment, where we examined the impact of using a large screen, an elision technique, connecting lines, and three-dimensional visualization on readability and recall of a set of graphical IS models. Our second prototype is able to produce 3D UML diagrams, to examine the model, and to show how diagrams are related to each other. We found out that visual integration techniques decrease the designers’ cognitive efforts to read and integrate diagrams and significantly reduce errors in both search and recall tasks, especially with respect to individuals with low spatial visualization ability. The research results improve our understanding of how design information is represented. It helps understanding relationships between different types of diagrams and between different levels of detail. This conclusion is applicable to other domains, where information is scattered in various visual forms and levels of detail.