Maria Francesca Costabile Dipartimento di Informatica, Università di Bari, Italy

1. Supporting Domain-Expert Users Maria Francesca Costabile Dipartimento di Informatica, Università di Bari, Italy costabile@di.uniba.it Pisa, 23-24 September 2002

2. Outline • Who we are • End-users • Domain-expert users • Characteristics of systems for domain-expert users • Software Shaping Workshops: environments for supporting domain-expert users development

3. Maria Francesca Costabile*Antonio Piccinno* Dip. Informatica Università di Bari, Italy Daniela Fogli*Piero Mussio*Dip. Elettronica per l'AutomazioneUniversità di Brescia, Italy Giuseppe Fresta ISTI "A. Faedo" CNR, Pisa, Italy *Pictorial Computing Laboratory University of Rome “La Sapienza” Italy

4. Pictorial Computing Laboratory • PCL started in 1993 at the University of Rome “La Sapienza” • Director: Stefano Levialdi • Main researchers: M.F. Costabile, Piero Mussio, P. Bottoni • Primary theme: Analysis and use of images in HCI • Formal theory of Visual Languages to model Visual Interaction • IEEE Symposia on Visual Languages, now Symposia on Human-Centric Computing • Journal on Visual Languages and Computer • Advanced Visual Interfaces (AVI) Conferences, AVI’04 in Bari

5. End-users • A user of an application program. Typically, the term means that the person is not a computer programmer. A person who uses a computer as part of their daily life or daily work, but is not interested in computers per se. Allen Cypher,“Watch What I Do: Programming by Demonstration“ • End-User Computing is “..the adoption and use of information technology by people outside the information system department, to develop software applications in support of organizational tasks” Brancheau and Brown, “The Management of End-User Computing: Status and Directions”, ACM Computing Surveys, 25 (4), 1993 • End-users are experts in a specific domain, who use computer systems to develop software applications

6. Domain-expert users • Domain-expert users: experts in a specific domain, not necessarily experts in computer science, who use computer environments to perform their daily tasks • They are responsible for the tasks accomplished through the system • They must understand the consequences of the system activity and must be in control of the interactive computation • Domain examples: • Medicine • Mechanical Engineer • Earth Science

7. The communicational gap • Users and designers adopt different reasoning strategies: • heuristic vs. algorithmic • examples, analogies vs. deductive abstract tools • concreteness vs. abstraction • Users are forced to express their problems in “computerese” Interactive systems are difficult to learn and use

8. Bridging the gap • Recognizing users as experts!!! in their field, not in CS • Recognizing that experts develop languages and notations to reason on problems and communicate solutions • Designing systems whichmakeabstract CS concepts concrete to users and allow users to follow their learning and reasoning strategies

9. The designer team Software Experts HCI Experts Domain Experts

10. User diversity within a domain • It depends not only on user skill, culture, knowledge, but also on specific abilities (physical/cognitve), tasks and context • “using the system changes the users, and as they change they will use the system in new ways” Nielsen, Usability Engineering • New uses of the system make the environment evolve and force to adapt the system to the evolved user and environment

11. Loop 2 Organizational context Technology User view of the task Loop 1 System Co-evolution Adapted from [Bourguin & al. 2001]

12. User tailoring • User diversity  user tailoring • Decoupling between pictorial and computational representations of concepts • PCL theory of visual sentences and model of visual interaction

13. screen computing reasoning materialization materialization interpretation interpretation PCL interaction model

14. characteristic structures • a characteristic structure (cs) is a set of pixels relevant to the interaction process Edit

15. attributed symbols • a set of css can be identified on an image i on the screen • the computer interprets every gesture with respect to the image i on the screen using a description d of it • d is a set of attributed symbols, each describing the computational meaning of a cs in i Bullet (b11, 279, 224, 18, blue, comp-action=‘link to a page’) tuple of properties

16. visual sentence • the relations between characteristic structures and attributed symbols are specified by 2 functions: • int: CSid (interpretation function) where CSi is the set of css in image i • mat: dCSi (materialization function) • a visual sentence (vs) is a triple • < i, d, < int, mat > > • a visual language VL is a set of vss

17. mat & int • A concept is materialised on the screen through a mat function that associates the attributed symbol describing the concept to a specific cs • The same concept can be materialised in another cs through a different mat function • int & mat allow user tailoring

18. Example in earth science domain • An Environmental Agency • Experts perform: • photo interpretation of satellite images to obtain medium and long term environmental forecast • organize the forecast results into reports and thematic maps for different communities of client experts (planner, decision maker, ...) • Two categories of experts arise: photo-interpreters and clerks, sharing similar notations but having different tasks to achieve

19. d int1 mat1 int & mat for the photo-interpreter

20. d int2 mat2 int & mat for the clerk

21. How to support domain-expert users • Design environmentsin which domain-expert users interact in their visual notations and with tools familiar to them

22. Software Shaping Workshops • Software environments as workshops in which users find and use virtual tools resembling their notations and habits and necessary to accomplish their activities • Analogy to real workshops, such as blacksmith or joiner workshops • More user categories  SSW hierarchy

23. System workshop B-SwEngineer generation generation generation System workshop B-AirManager System workshop B-GlacManager System workshop B-VegManager … … generation generation generation … … Application workshop B-glacier Application workshop B-monitore A 3-level Visual Workshop Hierarchy

24. The photo-interpreter environment The photo-interpreter obtains the spectral signature of the ‘ablation area’ using tools and notations familiar to her

25. The clerk environment The clerk obtains the glaciological parameters using the button ‘ablation area’ using tools and notations familiar to him

26. Conclusion • Software Shaping Workshops are computer environments aimed at supporting the activities of domain-expert users • Novel features of our approach are: 1) it is collaborative in that end-users, as domain experts, assume a responsibility in the design; 2) the SSWs are able to associate different pictorial representations to a same computational representation, and vice versa, thus permitting end-user tailoring; 3) end-users can perform their tasks interacting with the SSWs through interaction visual languages, which resemble their traditional notations and tools; 4) the system is evolutive and allows the domain-expert user to generate another SSW with new functionalities. • We are experimenting this apprach in various domains, using the software tool BANCO