ACM / Crossroads / Xrds10-1 / Learning to Use Virtual Team Collaboration to Solve Wicked Problems

Learning to Use Virtual Team Collaboration to Solve Wicked Problems

by Stephanie Cupp, Joel Foreman, S. Gievska-Krliu, and Rachelle S. Heller

Introduction

The focus of this paper is the ELEARNING RESEARCH PROJECT (hereafter referred to as the EProject), a project to investigate how virtual teams collaborate to solve highly complex or wicked problems. The EProject designed, constructed, and assessed a Web-based collaborative learning environment to support virtual teams of intelligence analysts. The mission of these geo-distributed and cross-disciplinary teams is to learn to collaborate in order to integrate knowledge from diverse domains and thereby produce solutions for wicked problems. Intended as an agent of organizational change, the EProject learning environment attempts to transcend the compartmentalized division of knowledge that limits inter-organizational cooperation and keeps cross-functional and interdisciplinary collaborative problem solving teams from forming.

EProject relies on three features: environment, context, and content. The environment used is a virtual team, the context is critical thinking, as demonstrated by collaborative problem solving, and the content is wicked problems and metacognition, or thinking about thinking. This combination presents the requirements and challenges faced by the researchers.

Background

The notion that the intelligence community could, and should, reform itself along the lines of a postmodern "corporate" model emerged in the 1990s. It is directly related to the transformation in business thinking and business practice produced by the proliferation and increasing sophistication of electronic networks. In 1996, the Commission on the Roles and Capabilities of the U.S. Intelligence Community (also known as "IC21") [10] reinforced this "corporate" model in its findings. The reports include a major commitment to the primacy of an electronic infostructure that would serve as the nervous system for the entire intelligence community and a criticism of that community's persistent bureaucratic and hierarchical mode of organization.

Virtual Intelligence Teams

The EProject collaborative learning environment supports the formation of virtual intelligence teams that are able to operate across traditional disciplinary boundaries and solve what would otherwise be intractable problems without this level of collaboration and cooperation. The scenario is intended to mimic an emergent crisis requiring the rapid formation of a virtual intelligence analysis team whose members might be drawn from several different intelligence agencies, each member specializing in a needed skill. The key features that define an extreme case "virtual intelligence team" are the urgency of its mission, its cross-agency and interdisciplinary make-up, its use of a collaborative tool suite [5], and the complexity of its task. These combine to produce a highly stressful work situation that places a premium on skillful interpersonal dynamics and web-based interaction.

Complex and Wicked Problems

The kinds of problems addressed by the EProject's virtual intelligence teams range from "complex" to "wicked." Complex problems have definitive and obtainable solutions whereas wicked problems open into a variety of alternative solution paths. None of these solution paths are "right" and each requires a different set of trade-offs to be considered. Schum [14] defines wicked problems as ones that cannot be easily defined such that all stakeholders agree on the problem, require complex judgments about the level of abstraction at which to define the problem, have no clear stopping rules, have better or worse solutions (not right and wrong ones), have no objective measure of success, have no given alternative solutions (these must be discovered), and often have strong moral, political, or professional dimensions.

Both complex and wicked problems require the same problem solving efforts including a "process that requires defining a search strategy, locating the needed resources, accessing and understanding the information that is found, interpreting the information, communicating the information, and evaluating conclusions in view of the original problem" [11].

The challenge of complex and wicked problems and their solutions is creating a heuristic by which those solving the problem can make sure they have reviewed all options, or as many as time permits. Typically, solving these problems requires the solvers to know a series of complex facts and extensive cases where similar situations were resolved. Problems pose the additional requirement that knowing and interpreting all the pertinent facts and cases is beyond the knowledge and experience of just one individual. Wicked and complex problems lend themselves well to team review, even though the team process brings additional challenges to problem solving.

Visualization and the Collaborative Learning Environment

Most of the collaborative problem-solving performed by the EProject virtual teams takes place within Web conferences using Mindjet's MindManager as a visualization tool and relies on a variety of communication channels to support its work. These include traditional email, web-based conferencing, and teleconferencing using a standard land-line phone link. MindManager supports team members as they create, link, and display visual knowledge maps. These maps assist in the formation and maintenance of the team's shared mental models, aligning the team so that all efforts are directed toward agreed-upon goals. Teams design these maps for easy consumption of complex networks of team knowledge, to provide information storage and retrieval through document and web links, and to provide the shared conceptual space within which teams at an intelligence agency can define and negotiate the meaning of these texts. Figure 1 shows a typical mind map for the first of the three learning modules. This particular map shows a partial view of the work involved for the team as they reflect on the entire process of the learning module.

Figure 1: Screen shot of partial mind-map: Learning Module 1.

Learning Modules

The Learning Modules were developed based on literature of critical thinking [2, 12, 15]. Skill sets larger than could be accommodated in a short learning experience were identified and then learning modules were built just long enough (20 hours) to test the ability to target a limited number of skills and produce an assured learning experience. The instructional design of the module begins with identifying those constructs that can define the three skill sets with a sufficient degree of specificity to insure that the team's performance can be validly measured or observed [13]. As a relatively new approach to virtual problem solving, visualization became by necessity a fourth skill set team members were required to develop concurrently with the primary learning objectives. The inventory provides criteria for observing evidence of skills. Critical thinking and meta-cognition can be observed indirectly, through "self-reporting," and as component skills for directly observable behaviors or performance, in creating MindManager maps.

For the skill of critical thinking, we consider the application, analysis, synthesis, and evaluation. Virtual teaming skill components include planning/coordination and communication. Visualization is represented by the concept of a mental model and the demonstrated use of various modeling tools. Metacognition is both self-monitored and self-regulated in the project.

Learning Module 1

Learning Module One (LM1) is designed to provide learners with background information, skill rehearsal, and measurable learning activities about the three aspects of the e-learning problem: the virtual intelligence team environment, the context of critical thinking as demonstrated by collaborative problem-solving, and the content of wicked problems and in meta-cognitive skills as demonstrated by thinking about thinking or reflecting on thinking. LM1 is an active, hands-on learning environment, designed to provide team members with the ability to read (observe), modify, and create learning artifacts (i.e., visualization maps). LM1 content is provided to learners in a series of layered concept maps. We believe that 'the medium is the message' and that by providing the material in a map, we underscore the importance of the concept map and we give students a series of opportunities to learn to read maps. We model the behavior we want them to adopt. This module, as with the others in the series, is directed at adult learners and is based on constructivist learning approaches. Students are given the materials to manipulate and background theory to describe the learning task. Within the four week, 20 hour module, they are free to move at their own pace within and with the group.

LM1 is developed to provide the learners with repeated experiences in visualization, including learning to read or follow material in concept maps, to modify them, and finally to produce their own maps. The specific critical thinking skills and learning experiences of LM1 track the three aspects of the e-learning environment. Students participate in a virtual team, identify a product role (specific to the wicked problem) they feel comfortable playing, assume a process role (i.e., recording observer, elaborator, encourager, or summarizer), and work together to create a group identity [1]. The current users of LM1 are modeling the actual target users of these learning modules, so they are asked to identify a product role from among four information assurance roles (threat analyst, technology analyst, cryptographer, or information analyst). One skill related to critical thinking is collaborative problem-solving. To that end, students read, modify, and create concept maps, use collaborative tools, share documents, compare and contrast collected information, hold synchronous and asynchronous discussions, and negotiate a team response to a wicked problem. The LM1 problem requires the team to create a policy statement for United States regulators concerning web page registration. The skills related to solving wicked or extremely complex problems require that the students review the characteristics of a wicked problem, collect data (facts and cases) on similar problems, compare and contrast collected data, and apply collaborative problem solving to a wicked problem. The problem space has its own demands. In LM1, the problem is intentionally accessible in that team members have an easy familiarity with the concepts, allowing them to develop the new skills related to visualization, team building, and collaboration. Finally, as reflection is key to learning, students are encouraged to reflect on each task or set of tasks.

Learning Module 2

In LM2, the students review and build on the core skill sets and cognitive abilities while working in information assurance. All weeks incorporate learning activities for individual and team work. The team work takes place online via Web conferencing made possible by MindManager and augmented by the telephone. During the four weeks, LM2 scaffolds the learning experience in Weeks 1 and 2 in order to maintain the focus on the limited skill set. Week 1 assures that the students thoroughly understand the goals of the learning module, the structure of the experience they will have, the tools they will use, the processes they will employ, and the deliverables they will be expected to produce. All of this information is provided on the "Orientation" map (Figure 2). The web-conference confirms that the team conforms to a shared mental model. The scaffolding, which falls away in Week 3, provides students with a clear sense of purpose, assures success in the early going, eliminates extraneous processes that could impede progress toward the targeting learning goals, and maintains focus on those goals. The momentum accelerates at that time and the students are gradually released into an open practice field where they have the opportunity to apply the skills exercised in previous weeks.

Figure 2: Screen shot of "Orientation" map (1 of 5) from Learning Module 2.

Each team member individually complete a "virtual team skills checklist" and "meta-cognitive log." The "after action review" conducted at the end of each week occurs online and is managed by a facilitator who helps the students reflect upon the collaborative activities, identify lessons learned, and prepare to apply them in future work.

In Week 2, the students receive the information assurance problem they must solve and they build initial and goal state maps as the first step in the critical thinking and collaborative problem solving process. They produce individual maps, then combine them in pairs, and finally reach a team map. The process aggregates insights, encourages dialogue and negotiation, and promotes comparison and contrast and other critical thinking skills. These ideas are built using means-end analysis [9]. The gap between the initial and goal states becomes the problem space [6], in which team members attempt to link available evidence or conceptual frameworks with new knowledge obtained through meta-cognitive reflection, group reasoning, or sleuthing. They receive a variety of documents that hold the solution to the problem and must plan how to use the five hours of the week to divide the work and coordinate their activities that will lead to the development of their solution to the problem.

This team-structured problem solving work continues into Week 4 and concludes with their construction of a map that outlines their proposed solution. A learning module wrap up follows with each team member completing "360-degree evaluations" of himself/herself and the other members of the team. The final "after action review" reflects upon the entire learning experience and reinforces major points of learning, and the module closes with a web conference during which the author of the information assurance problem debriefs the student team.

Assessment

The assessment employed various tools: a pre- and post-work interview, a review of artifacts (knowledge maps, email, threaded discussions, and other documents) created during the learning modules, a skills-based test on the ability to use the tools and read, modify, and create MindManager knowledge maps, review and analysis of web conference transcripts, review and analysis of web conference observations, and a collection and analysis of survey data.

An Informal Assessment of LM1

LM1 assessment is based on a pre-learning module interview to obtain baseline information about the participants. The team members were chosen from among graduate students in computer science at the participating universities. Team members reported that they had very limited experience with distance education, but that their impression was that it was a "good idea." In fact, they thought that a positive aspect of the project would be the experience of distance learning. All had worked on team projects in their classes at one time or another, but none had participated in a truly virtual team. In their previous experiences, when they were able to establish team cohesion, they did so based on class participation and the need to earn a grade for the course. Most of their interaction was face-to-face or follow ups to face-to-face class meetings, and communication was augmented with email and a few phone messages.

No team member had used MindManager before, nor did they recognize the term "visualization maps." While they report taking notes in class and reading (and rereading) the textual material, they did not relate to the question about how they reflected on their learning - in fact, one team member reported, "I don't think about how or what I learn at all, I just learn." However, all reported an easy familiarity with general computer processes including creating, storing, and retrieving files, sending email with attachments, and surfing the web using search tools. A few had tried chat rooms and none reported playing multi-player games on the Internet. They relied on the Internet and course material for problem-solving support and did not use the library or other outside resources. They reported preferring email communication for friends who are far away or for short messages for friends and co-workers alike. They relied on the telephone for catch-up visits with friends.

Team members did not relate to the term "wicked problem" and their experience on team projects had only required them to work on a problem for which every team member had similar capabilities and experiences. These class problems were big problems but not particularly complex or wicked, according to our definition. The team goals were defined by the teacher, and team activities were directed by the need to complete the project for a grade.

After LM1, the team members were interviewed again and they were required to demonstrate their ability to use the MindManager tool as well as the file storage and chat tools. They reported enjoying the experience but felt they did not have sufficient time during the module to complete the tasks to their own satisfaction. All users were comfortable using the communication tool to store and retrieve files, to see if other team members were online for chat, and to create chat rooms for discussion. Also, all were able to use the tool to read, create, manipulate, store, and share maps. The ability to initiate a map conference, which is different from joining a conference and sharing a map, posed the only problem during the interview. However, the assumption is that the firewall instructions were not repeated carefully because three of four users had reinstalled MindManager onto a new computer after LM1 for this interview. For those users that were not able to initiate a conference, the maps used during the interview were stored into the files tool for review.

Additional interviews after LM1 indicated that the most difficult aspect of the learning module was dealing with the actual technology. Installing the software presented challenges to some who had firewalls in place. Though LM1 urged the team members to use the tutorials provided by the software, it was not until they actually had to use the tool to engage in the learning module that they realized that their understanding of the tool was limited. They needed to practice using the tool. The team reported other difficulties in sharing maps and in reading the text notations provided on some of the links. In fact, one student who had to travel during the learning module sequence printed out the text notations and only then did the team member report that he or she understood the text sections and used them regularly after that. Some of the other team members did not realize that there were directions, rationales, and references in the text sections.

LM1 was designed as a constructivist learning environment. While the branches were numbered (Figure 3), team members reported they felt adrift and "didn't know what they were supposed to do." It was not immediately clear to them that they were in charge of their own learning and could decide when to meet, what to do, when to do it, and how much of the tasks to complete . Once they did, they made reasonable strides. They reported "enjoying the freedom" to work on their own and the serendipity of discovering that the team-building map that required them to introduce themselves to each other with a 'curious fact' about themselves broke the ice and the teams moved forward after that.

Figure 3: First map of learning module. Note the items that are numbered for suggested priority (1, 2, 3).

The teams formed a democracy. Team members did engage in process roles and each took the role they felt fit their experience or personality. They reported that the 'encourager' was perfect for Team Member A; that person had a nice way of keeping them on target. The 'summarizer' was well suited to producing syntheses of the meetings and reports on directions yet to be taken. The product roles were more difficult to assume. These roles were specific to intelligence agency team members and were unfamiliar to the LM1 team members. Also, the information provided in LM1 was not precise enough for the team members to distinguish between the first three roles. Therefore, when it came to negotiating the final policy statement from among the group members, they felt they had no way to identify the positions that each product role member would take in the real world.

The team reported that, as a team, they came to develop their own communication style. They relied on email for short bursts of information and phone conversations for long work sessions. While they tried chat rooms, the technology problems, including lost lines and frozen computers, reduced the usefulness of that tool. Email proved to be good for scheduling meetings, and for sending quick notes and queries. Multi-way phone conferences were well suited for supporting work sessions.

An Informal Assessment of LM2

The learning outcomes for LM2 are intended to promote collaboration, critical thinking, and metacognition within a problem-based learning scenario. A rubric for assessing the team members' success or failure to create viable maps is based on the four competencies or types of knowledge identified by Mayer, Larkin, and Kadane [7] as essential for solving word problems. This taxonomy suggests that a learner has thorough knowledge of a subject, accurate but incomplete knowledge, some knowledge, or an incomplete set of ideas about a topic. Moreover, a learner can have thorough knowledge in the linguistic or algorithmic area, but not at the schematic or strategic domain. Depending on the material incorporated in the learner's maps, assessments can be made concerning these areas. For example, if a learner has complete algorithmic knowledge, then the map should indicate "significant constraints on the developing solution," but if the learner has some 'holes' in the learning, then the map might only show some alternative solutions.

An interview of the team was conducted at the conclusion of LM2. Participants expressed some concern that the four-point Likert scale used to measure various virtual teaming skills was too blunt an instrument and did not capture some of the subtle detail of their performance. Team members identified communication as the most important of the sub-skill sets and indicated that the low bandwidth of virtual teaming could often lead to misinterpretation. All participants agreed that they had some trouble remembering specifics when responding to questions on the reflection log that ask for information about the previous weeks activities.

The students found the learning module to be clear, directive, easy to understand, well laid out, and easy to follow. These important claims are supported by the students' ability generally to perform properly and on time. Three of the four participants in LM2 reported that they overcame initial disorientation with the MindManager environment and "now prefer mapping when working with a group." One participant complained that "some aspects of mapping were not intuitive" and the maps "took some time to interpret." The team was unanimous in its opinion that the five hours it set aside each week for the module's activities were not enough. It was suggested that more time was necessary for discussion and analysis of the problem.

Although the LM2 design emphasized simplicity, the module still produces too much of a cognitive burden within the five hour/week time constraint placed on the users. Two pieces of evidence support this claim. First, all the students said they would have liked much more time to perform their work. Second, they did not consistently act upon all of the directions in the module - especially those relating to the completion and submission of the "virtual skills checklist" and the "reflection log." The students were very engaged by the information assurance problem and worked hard to come up with a solution. They came up with a number of possible solutions, but were frustrated to not have a definitive and correct solution.

The structure of the virtual team was a positive experience in that the students reported they felt the need to be very disciplined and task-oriented. Because of the time constraints, the web conferences had to be well organized and they were. On the other hand, the students would have liked more time for online socializing.

The students found that the facilitated "after action reviews" played an important role inclarifying confusions and misconceptions that had developed during any given week. As elsewhere, the students felt they did not have enough time to prepare for the reviews. They would have liked more time to focus on specific weekly learning goals. They would have liked to see the role of the after action review clarified and made more explicit and follow through with the goal setting and application of lessons learned.

Changes Based on Assessment

As EProject continues and Learning Module 3 is developed, LM1 will be revised based on our assessments. The major revisions include a more detailed welcoming note that describes the constructivist learning environment and gives the team members permission to take their own learning direction. LM1 will have additional time devoted to tutorials on MindManager. The product roles, which will be different in LM3 from those in these modules, will be more deeply defined. Team members will be given thorough descriptions of the product roles and examples of the type of response each of the product roles would take in a similar situation. These tasks are not intended as learning objectives of the module, so preparation may help "jump-start" the learning that will take place in the module.

Help is provided in map form, consistent with the visualization technique used throughout the module. To improve the opportunity for help documents to assist the users, the help maps will be duplicated in a lateral document and stored in the file tool. Also, to assist the users in speedy recovery, help documentation will have more directed sources within the tutorials for specific cases. More help is needed for users to prepare before the learning module experience begins, especially in describing the constructivist learning environment and giving the team members permission to take charge of their own learning direction. Additionally, users need comfort with solving complex or wicked problems and ideas on how to confront inexact solutions. A tweaking of the information assurance problem may solve this problem. The trick is to give the students just the right amount of information. Unfortunately, the "right amount" (i.e., that which is enough to cause several hours of focused collaboration and results in a right solution) is difficult to quantify exactly.

The users expressed concern over delay caused by technical problems, such as connection failure or computers freezing. Taking advantage of the extra time to prepare the technical tools will remove any start-up frustration when the module begins, allowing the users to focus on the project. Future modules will provide suggestions on how to continue the work in another manner if such issues arise, in order that the users may continue their discovery without delay. The intention is that as the users become more familiar with tools available to them, they will be able to improvise on their own.

Suggestions will be provided in the maps based on these users' experience for ease of use. For example, in the first map, a mention of how one user found it helpful to print all of the notes in the map will be noted. Another suggestion will be to schedule a test conference so that each member can master the conferencing procedure before topic meetings are scheduled.

Establishing team identity is an important aspect of the virtual teaming experience. A redesign of LM2 should enable the social interactions that set the stage for more intense task-focused engagements.

Lessons Learned

While the analysis and evaluation of EProject is primarily directed toward the continual improvement of the project, guidelines can be adopted for other projects involving collaboration and virtual teams. Many of the changes for future modules are relevant to general collaborative work.

Time

The most basic lesson relates to time. Collaborative projects need time to develop. The team-building activities proved worthwhile in developing the team cohesion and decreasing the time necessary for team building. Time needs to be set aside for a team to gain a reliable knowledge of the tools. Time for reflection has to be reserved and activities structured around reflection must be established. The team process roles also enabled the teams to build cohesion easily and early.

Medium

The next lesson relates to the medium and the message for the collaboration. It is suggested that because the presentation of the materials and the workspace within which to accomplish them are consistent, the problem space and the solution space both should rely on the same tool and presentation. Visualization is not an intuitive process and needs to be scaffolded. Virtual teams can benefit from the visualizations, but they need to be provided with examples of good visualizations upon which to model new ones. A series of partially completed maps provides team members with an opportunity to adapt and adopt the maps to their particular task.

Assessment

The last lesson relates to the vagueness of the solution assessment for wicked and complex problems. As products of our educational system where there is always an answer, users do not feel comfortable with the solutions. More work needs to be done to help users feel that the solutions they develop within the resource framework are acceptable and worthwhile. It should almost go without saying that communication is a key tool for virtual teams. As many communication paths as possible should be available to the team members. While different teams may choose differently, our experience shows that for simple communications including agenda-setting for more detailed meetings, email was the most efficient. Additionally, teleconferences are critical for negotiation and discussion.

Acknowledgments

This project was conducted jointly by a team from George Mason and George Washington Universities, Washington DC. Technical help and support was provided to E-Project by graduate students S. Gievska-Krliu at George Washington University and Luciano Viera at George Mason University. Additional efforts were provided by Jerry Drake, who has a MEd in instructional technology from George Mason University and is presently pursuing a PhD at Mason's Graduate School of Education.

References

1

Baum, J. A. C. Being a Valuable Team Member, <http://www.rotman.utoronto.ca/~baum/mgt2005/valuable.html>.

2

Chandler, K. "Metacognition: Its components, development, and relationship to problem solving performance"

3

Davis, J. P. Fletcher, A. J. P, Shenton, W. W., Han, B., and Pang, J. Complex Problems with Incomplete Evidence: Modeling for Decision Making. 6th International Conference on Greenhouse Gas Control Technologies. October 2002.

4

Gagne, R. M., and Briggs, L. J. (1974). Principles of Instructional Design. New York: Holt, Reinhart and Winston.

5

Heller, R. S., Foreman, J., Gievska-Krliu1, S., and Cupp, S. Collaborative Teams Solve Wicked Problems: Theory into Practice, EdMedia 2003, Honolulu HI.

6

Holyoak, K .J. (1995). "Problem Solving." In D. N. Osherson (Gen. Ed), E. E. Smith, & D. N. Osherson (Vol. Eds.), Thinking: Vol. 3. An Invitation to Cognitive Science (2nd ed., pp. 267-296). London: MIT Press.

7

Mayer, R. E., Larkin, J. H., & Kadane, J. B. (1984). A Cognitive Analysis of Mathematical Problem-solving Ability. In R. J. Sternberg (Ed.), Advances in the Psychology of Human Intelligence. Hillsdale, NJ: Lawrence Erlbaum.

8

Mindjet, <http://www.mindjet.com>

9

Newell, A., and Simon, H. A. (1972). Human Problem Solving. Englewood Cliffs, NJ: Prentice-Hall.

10

"IC21: The Intelligence Community in the 21st Century," 1996. Staff Study, Permanent Select Committee on Intelligence, House of Representatives. <http://www.access.gpo.gov/congress/house/intel/ic21/ic21_toc.html> (27 February 2003).

11

OSLA Policy on the School-Library Information Centre and the Role of the Teacher-Librarian, 1996.

12

Rothbart, D. (2002). "Critical Thinking and Visualization." George Mason Technical Paper.

13

Shepard, L. E. (1993). Evaluating test validity. Review of Research in Education, 19, 405-450.

14

Shum, S. (1997). "Representing Hard-to-Formalise, Contextualised, Multidisciplinary, Organisational Knowledge." AAAI Spring Symposium on Artificial Intelligence in Knowledge Management, 24-26 March 1997. <http://kmi.open.ac.uk/people/sbs/org-knowledge/aikm97/sbs-paper2.html> (27 February 2003).

15

Tarmy, G. (2002) "Virtual Team Skills: A Review of the Literature." George Mason Technical Paper.

---

Biographies

Rachelle S Heller (sheller@gwu.edu) is a professor of computer science at George Washington University. She is also a co-editor of Computers & Education: An International Journal. Her research interests are in educational uses of computers and in interactive multimedia.

Joel Foreman (jforeman@gmu.edu) is an Associate Professor in the English Department at George Mason University. He has been doing research on virtual and collaborative learning since 1996.

Stephanie Cupp (scupp@gwu.edu) is an MS student in computer science at George Washington University. She received her BS in computer science at Kennesaw State University. Her research interests are in human-computer interaction, especially in the areas of web and software design.

Jerry Drake (gdrake@gmu.edu) has a MEd in instructional technology from George Mason University and is presently pursuing a PhD at George Mason's Graduate School of Education.

Sonja Gievska-Krliu (sonjag@gwu.edu) is a DSc student at the George Washington University. She received her BS at the University of Kiril & Metodij, Skopje, Macedonia and her MS at the University of Zagreb, Croatia. Her current research interest is focussed on context-sensitive user interfaces.