The digital transformation leads to software systems pervading almost all spheres of private and professional life. To ensure that these software systems are designed and successfully implemented as needed, intensive collaboration is essential in the key roles in software projects, in particular for the roles of product owner, user experience designer, as well as software engineer. The collaboration of people with usually different levels of IT-savviness requires the appropriate skills of those involved, which are also called Future Skills. Computational thinking is an important skill for everyone involved in software projects, no matter which role they are in. We describe an interdisciplinary tool-based teaching-and-learning program where we build virtual voice-based assistants (voice apps for Amazon Alexa) in interdisciplinary student teams to train computational thinking and collaboration skills. A first competency test validates the effectiveness of our approach.
In current digitalization initiatives, there is a lot of
discussion on how to increase graduation numbers
in software engineering related study programs in
order to have more skilled people driving the ongoing
digital transformation. In this discussion, however,
we often forget that digitalization is always related
to an application domain. The digital transformation
benefits strongly if software-related skills are
strengthened not only for the core software engineering roles,
but also for less-technical roles in software projects
1With ’Future Skills’ we refer to ’competencies’ required by
university graduates across all majors in the coming years. These
competencies are necessary to meet digital requirements as they
are currently expected in business and society (cf.
Software products can only be successful if the key roles work hand in hand in software projects: product owners with their knowledge of the application domain and product vision, user experience designers who guide human-computer interaction, and software engineers being responsible for software implementation. The skills and competencies related to these roles are essential in successful projects. They are required to successfully meet the challenges of digital transformation.
How do we best train the talents for the ongoing
digital transformation, and what exactly do future
employees need to learn? Stifterverband, a joint initiative
of German organizations, has published a discussion
paper in September 2018 together with McKinsey &
Company where they address three core competency
categories driving the digitalization
With changes in the job portfolios and new forms of work, there has been an expected shortage of qualified people for some time now. In particular, the job market will be increasingly dominated by job profiles that are heavily related to software engineering. Accordingly, a lot of effort is spent on increasing the number of graduates of software engineering related degree programs, in order to increase the number of software engineers in the market and thus to meet the growing demand for qualified specialists.
At the same time, it is essential to strengthen everybody’s digital and non-digital key competencies. Digitalization is not just an IT issue. Digitalization is inherently linked to the digital transformation and thus the creation of digital automation or digital assistance systems in an application domain. Therefore, it is just as important for employees and students of all application domains to acquire software engineering competencies.
Stifterverband
How can we integrate the development of these competencies into study programs? Curricula are often overfull and the amount of available (and often essential) knowledge is increasing in almost every application domain. Solutions have to be found to systematically integrate key competencies required for the digital transformation into the teaching-andlearning processes without weakening the core foundation in the respective application domains.
Project-based learning is a successful instructional
learning format that has been proven to systematically
strengthen some of the required future skills (Bell,
2010). Interdisciplinary project-based learning is even
more effective, as team diversity is an additional
success factor for creative, goal-oriented collaboration in
addition to the future skills mentioned a
Wing (2006) introduced the importance of computational thinking for computer science-related tasks 12 years ago. She defines computational thinking as the interplay of decomposition and abstraction and recommends strengthening computational thinking in all study programs.
The ability of abstract thinking, in turn, has long
been recognized as a key competency of many
technical disciplines, especially for computer science
The central element of computational thinking is a
problem-oriented (as opposed to a solution-oriented)
approach
In teaching-and-learning practice, it can be
observed that not all students have a sufficient level
of abstraction and computational thinking to be able
to cope with the study program requirements. This is
especially true for students of subjects related to
computer science. Accordingly, various approaches have
been developed to systematically strengthen these
abilities (Hazzan and Kramer, 2007;
In this work, we develop and apply a teaching-andlearning program for promoting computational thinking in students – and, as an essential basis for this, for fostering students’ static and dynamic abstraction competency. To this end, we develop a teaching-andlearning program for interdisciplinary, project-based learning that addresses computer science students as well as non-IT students. Our concept takes into account our students’ prior knowledge as well as their individual learning requirements.
In our teaching-and-learning program, we create an easy introduction to digital projects for interdisciplinary student teams. To achieve this, we use a tool-chain for creating voice-based virtual assistants (such as Amazon Alexa), as well as by using Github. Over the last years, the usability of many tools for creating software systems has evolved and improved significantly. Digital tools in the context of software engineering are nowadays no longer editors that are specifically tailored to computer nerds. Rather, nowadays they often are web-based interactive tools that are fun to use and that guarantee good results even for people without IT- and software engineering skills.
Our student teams include tourism majors and computer science students. Github is used as a source code repository by computer science students, as well as a ticketing and project communication tool for all students. We furthermore use the Github project board as a virtual agile board. In this way, we enable all students to make an active, creative contribution in a digital software project even without programming knowledge.
The non-IT students (i.e. students of an application domain) are either in the role of product owner or user experience designer. The role of product owner includes gathering and aggregating the users’ needs and desires which requires static abstraction capabilities. The user experience designer role focuses on reflecting what users expect. This definitely requires consideration of dynamic processes. In parallel, computer science students deepen their experience in the role of software engineer by using new technologies for the implementation of voice-based assistants.
We use pseudonymized pre- and post-tests to analyze to what extent our interdisciplinary tool-based teaching-and-learning program actually fosters the addressed competencies of static and dynamic abstraction in the students.
Voice-based virtual assistants (voice apps) are currently being massively pushed by major software companies. Examples are Amazon Alexa, Google Assistant, Cortana, and many more. In particular, Amazon and Google provide well-designed, web-based tools that developers can use to create new voice apps with their cloud offerings. These tools and cloud offerings are available free of charge for educational use. In addition, the tools are so sophisticated that it’s fun to play with them. In just a few minutes, cool voice-userinterfaces can be created without previous knowledge, so first success can be achieved quickly. Figure 3 shows the Alexa Developer Console, a tool for creating voice apps for Amazon Alexa.
To develop an application for voice-based assistants, the development team must design the Voice User Interface (Voice UI) and implement the business logic. A voice UI must be structured in such a way that the dialogue appears natural to the users. At the same time, the dialogue must be designed so that the goals expressed by the users or the intentions behind them can be clearly identified and assigned to the implemented business logic. In the jargon of voice UIs, these intentions are called intents. Each intent must clearly invoke a feature of the implemented business logic.
Figure 4 shows an example dialog with Alexa for an application called the "joke-of-the-day". A person starts the voice app and gets told one joke. More jokes can be requested. If no more joke is desired, the voice app closes.
Structuring a dialogue into intents strengthens both static and dynamic abstraction abilities.
The identification of the individual intents in a dialogue requires a static abstraction. For example, the joke-of-the-day application is structured into the following intents and the resulting dialog steps: welcome joke closing
The individual dialog steps or intents must then be arranged in a meaningful sequence (the dynamic behavior). This process is rather simple for the jokeof-the-day application (see Figure 5).
The design of a dialog for a voice-based virtual assistant is thus well suited to train both static and dynamic abstraction skills, and to guide students towards computational thinking. Therefore, we use dialog design and implementation for a voice-based assistant as a task for an interdisciplinary tool-based teaching-and-learning program to strengthen computational thinking.
Our didactic concept for the promotion of computational thinking and communicative future skills structures the learning process into several phases (see Table 1). Initially, the students are taught core concepts in non-interdisciplinary groups.
More precisely, the computer science students first learn the technical basics of voice-based assistants, and are introduced to tools for designing and implementing them. Furthermore, computer science students learn requirements engineering. Having that down their belts, these students are well equipped to Task
Computer Science Students
Students of Application Domain Introduction to voicebased assistants
Examples and tutorial for creating a first voice app
Examples and simulation of dialogues between two partners Idea for voice app and its structure
Invocation, intents, slots
Generating ideas and defining MVP Concept for Voice App
Requirements engineering
Specification of a first dialogue Using tools
Alexa Developer Console, Github
Invocable, Alexa Developer Console, Github 1st Pitch: Students of application domain pitch their ideas to find computer science students for their developer team. 1st Sprint 2nd Sprint
2nd Pitch: Voice app demos by computer science students.
Refining the dialogues
improvements take on the role of software engineer in the interdisciplinary teams that are formed later on.
The task of the application domain students is to develop an idea for a voice-based assistant (in this case an Alexa voice app). For this idea, they then define the dialogue between the user and the voicebased assistant required for a Minimum Viable Product (MVP). This dialog thus contains at least those steps and procedures that are necessary for the minimal functional implementation of the idea. It is important to break down the dialogue into short, simple steps and to structure it. The complexity of creating the voice app is significantly greater than the example of the "joke-of-the-day" shown in Figure 4. Suitable ideas for voice apps are (quiz) games, guides, or useful assistants.
The dialogue is tested and tuned by the students of the application domain. Then, using Invocable2, the students of the application domain create a first interactive but hard-coded prototype of the voice app.
The interdisciplinary collaboration in the teams begins when the students of the application domain present their ideas to the computer science students. The students of the application domain pitch the prototypes of their voice-based assistants to computer science students and try to motivate them to join their 2Invocable: https://www.invocable.com development team. Following these pitches, mixed teams are formed each consisting of computer science students and students of the application domain.
Within the interdisciplinary teams, the computer science students give feedback to the students of the application domain on the design of the dialog that underlies the respective prototype. Based on this, the voice user interface is subsequently refined together. The computer science students take their "natural role" as software engineer in the interdisciplinary project. The application domain students, on the other hand, are both product owner and user experience designer. So they design the interaction between the user and the voice-based assistant in such a way that this interaction is technically meaningful and needs-based from their own perspective. So far in our didactic concept, user experience design is not explicitly taught due to lack of time. However, it would be desirable to improve this in the future.
Computer science students implement the back-end, while students of the application domain are responsible for the front-end in the implementation phase of the voice-based assistant. A simple form of Scrum is useful for organizing the development process into sprints, thus structuring the semester process. Github repositories, including the integrated project boards and the integrated ticketing system (Github issues), support team collaboration through an appropriate tooling infrastructure.
This teaching-and-learning program was used for the first time in the winter semester 2018/19 at Munich University of Applied Sciences. Four computer science students (3rd semester) and two to three students of tourism management (6th semester) form a mixed team for the pilot run of our program.
Even if static and dynamic abstraction are not explicitly addressed, all students in this one-semester interdisciplinary tool-based program train their static and dynamic abstraction abilities by structuring and specifying the dialogue between a person and the voice-based assistant in such a way that a corresponding Alexa voice app can be built. All students (including application domain students) work with the Alexa Developer Console and Invocable tools. The use of tools enables all students to work with a working interactive voice app. The working prototype provides rapid feedback so that in particular the students of the application domain can immediately check their dialogue structuring.
From our perspective, this interdisciplinary toolbased teaching-and-learning setting is well suited to foster our students’ abstraction skill and more effective in this area than regular class exercises. Furthermore, standardized processes, such as completing Github Issues in team communication, help to structure collaboration.
As we were interested in investigating the development of our students’ computational thinking skills during the interdisciplinary tool-based teaching- and learning program, students were requested to work on a competency test at the beginning (week 2) and at the end of the semester (week 11). The test covered two facets of computational thinking, namely static and dynamic abstract thinking processes.
The test was taken by two groups of students, all of which were working on voice-based virtual assistants in a project based way. The first group are computer science students and students of the application domain who worked together in interdisciplinary teams. The other group, our control group, consisted purely of computer science students.
All these students were asked to solve two tasks that both form our competency test and that require static and dynamic abstract thinking processes, respectively. Using pseudonyms allowed us to match the students’ pre- and post-tests and thus, to analyze their individual learning outcomes. (Regarding the tourismmanagement students’ test performances, we will discuss the results in the next section “Coding scheme of the abstract thinking test & first results”. The test performances of the computer science students have also been analyzed, but will not be presented in this paper.)
Our hypothesis is that the computational thinking skills of students that actively participated in our interdisciplinary tool-based program increased significantly. More specifically, we assume that all students benefit from developing and implementing their innovative ideas on Alexa voice apps, as both the design and the creation of voice-user-interfaces require static and dynamic abstract thinking processes. Note that even though we expect that students working in interdisciplinary teams strongly train their collaboration skills during the course, we have not explicitly assessed theses skills in a test.
The first task of our competency test includes a menu of 22 different coffee specialties and requests the students to teach a new barista about the coffee recipes. The menu contains pictures and ingredient lists for each coffee specialty. The challenge is to structure the various coffees and their ingredients in such a way that a new barista can quickly grasp and learn them. In the second task, students are asked to generalize the abstraction process they applied when solving the first task.
To successfully accomplish the second task, the participants first have to become aware of their own actions, identify and structure those processes, and derive a procedure that would be transferable to similar tasks. Therefore, they have to dynamically abstract from their own approaches and accurately document their solutions. From our perspective, task 1 mainly deals with static abstract thinking processes, whereas task 2 covers dynamic abstract thinking processes.
Students are allowed 20 minutes for working on the first task and 15 minutes for completing the second task. Once the students begin working on task two, they are not allowed to use the documents they have generated in task one. In winter semester 2018/2019, 18 tourism students and 54 computer science students participated in the computational thinking competency test.
To analyze the competency test of computational thinking skills, a comprehensive coding scheme was developed incorporating five criteria to evaluate static abstract thinking processes (S1-S5) and three criteria to measure dynamic abstract thinking processes (D1-D3). We defined two additional criteria operationalizing one’s ability for self-reflection (R1) and for identifying the requirements needed (R2). All criteria differentiate between four levels of competency (outstanding, good, satisfactory, not yet satisfactory). Score 1 characterizes the lowest level of competency, score 4 the highest one.
We attempted to capture static abstract thinking processes by evaluating students’ performances along these four criteria:
Criterion S1: developing categories that ideally
simplify the given representations (of coffee
specialties) and structuring ingredients;
Levels of competency may be described as
’unrefined’, ’put together’, ’structured’ and
’organized’
Criterion S2: identifying and parameterizing attributes. Criterion S3: presenting categories in a structurally-sound way, e. g. as UML-diagrams. Criterion S4: using formal notations in a logically consistent way.
Criterion S5: recognizing familiar structures and realizing that the structures are coherent in a given situation.
Dynamic abstract thinking processes are operationalized by defining these three criteria:
Criterion D1: identifying a coherent chain of processes.
Criterion D2: presenting a coherent chain of processes in a structurally sound way.
Criterion D3: relating the identified chain of processes to one’s own solution, e. g. using one’s own approach as a basis for generalizing and inferring appropriate processes.
We considered two additional criteria, characterizing one’s ability for self-reflection as well as metacognitive thinking skills:
Criterion R1: identifying errors and reflecting one’s own approach.
Criterion R2: taking the users’ preferences into account, as well as the requirements.
First results indicate that the computer science students demonstrate a significantly higher initial level of static abstract thinking skills (mean-value M=2.04; standard-derivation SD=0.49), in comparison to the tourism majors (M=1.77; SD=0.46) (t=2.119; df=70; p=0.030).
Regarding the students’ initial dynamic abstract thinking skills, we found no statistically significant differences between the performances of computer science students (M=2.12; SD=0.51) and tourism students (M=2.28; SD=0.43; whereby U=384.00; p=0.210). It seems that tourism students have reached a slightly higher level of dynamic abstract thinking skills than computer science students.
Furthermore, the results indicate that computer science students (M = 2.07, SD = 0.78) and tourism students (M = 1.81, SD = 0.73) do not differ significantly in their self-reflection skills (U = 384.00, p = 0.210). We observe only a slight tendency in favor of the computer science students.
Overall, the competency test gives some insight that most of the students – computer science students (3rd semester) as well as tourism students (6th semester) – struggle with demonstrating their static and dynamic abstract thinking skills. According to Wing (2006) and based on our results, we strongly recommend the integration of interdisciplinary teaching- and learningprograms in students’ curricula to foster students’ computational thinking skills. Regarding students’ reflection skills, it seems to be important to encourage students to reflect their own approaches and to think about the requirements needed.
The results of the pre- and post-tests provide empirical evidence that the tourism students’ static abstract thinking skills development can be characterized by a significant increase (t = -3.986, p = 0.001) supporting our hypothesis. After actively participating in the interdisciplinary tool-based teaching-and-learning program (M = 2.21, SD = 0.48) the tourism students score significantly better in terms of their static abstract thinking skills than before their participation in this program (M = 1.77, SD = 0.46). The effect size by Cohen (1992) is around r = 0.695, thus indicating a strong effect. Regarding the tourism students’ dynamic abstract thinking skills, no significant effect can be reported (t = -1,475, p = 0.159). However, there is also a tendency towards an increase of dynamic abstract thinking skills (M = 2.24, SD = 0.44). Our hypothesis that tourism students improve their computational thinking skills during the interdisciplinary tool-based program can be confirmed.
Project-based teaching presents many challenges to
teachers
Make sure that the project tasks cover the planned learning content to the desired extent and adequately demand and strengthen the competencies to be acquired.
Build social interaction structures within the project teams that allow a balanced distribution of roles and tasks.
Create a good (i.e. applicable) schedule.
128 students in computer science and tourism management were in the pilot group in the winter semester 2018/19. We combined a software engineering module and a module for digital marketing and management. Both modules have their own learning objectives and content. In addition to the learning objectives of these respective modules, additional learning objectives of the interdisciplinary tool-based teachingand-learning program include the increase in static and dynamic abstraction abilities mentioned in this paper as well as the improvement of collaboration and communication skills in digital projects. Therefore, instructors need to encourage students in their learning process. At the same time, instructors have to take care to not overburden their students.
We solve the challenges of the different learning objectives and content of the initial modules by running some part of the interdisciplinary collaboration between the students in a purely virtual way, using the cloud-based Alexa Developer Console as well as Github and Github issues. The classroom events where both computer science students and students of the application domain are together are the three pitches which are highlighted in gray in Table 1, i.e. the pitch of the prototypes by the students of the application domain, the pitch of the voice app demos by the computer science students, and finally the joint pitch during the final presentation. A mix of virtual and physical collaboration creates enough space for both teaching and learning sessions to accommodate the specific contents of the respective modules and the corresponding competencies according to the definition of learning goals.
The interdisciplinary project was a lot of fun for everyone involved. For that reason alone, it is highly recommended to repeat the project. The use of new web-based development tools was well received by all participating students independent of their field of study.
The computer science students were motivated primarily by the fact that new voice technologies were used in the context of this project. In turn, tourism students have grown into the role of product owner during the project. Furthermore, by using the development tools, they were in able to increase their competency in using web-based tools. They also liked to creatively integrate sound effects into the Alexa voice apps.
It is important for all students to develop and to
strengthen their ability of computational thinking in
order to meet the requirements of the digital
transformation. As a basis, it is helpful that the students first
develop the skills for static and dynamic abstraction
as these are a basic building block of the ability of
computational thinking. Computational thinking is
one of the Future Skills
We strengthen computational thinking through interdisciplinary, tool- and project-based learning. As a project topic and work context, we select the design and implementation of voice-based digital assistants (so-called voice apps). The students are encouraged to use static and dynamic abstraction for the specification of the dialogue between a human and a voice-based assistant.
A competency test was developed in order to measure the effectiveness of our approach. The test was run at the beginning and at the end of the semester. (The evaluation of the post-test is not yet completed.) The pseudonymized test results are used to determine the extent to which the students were able to improve their abilities of static and dynamic abstraction during the program.
More tests are required for a more detailed analysis of future skills through interdisciplinary tool- and project-based learning in digital projects. Accordingly, we plan to improve and further expand our approach so that additional competencies are targeted and the effects are captured by additional measuring instruments.