Application of domain engineering to generate customized information dashboards
- 1. Application of domain engineering to generate customized information dashboards HCI 2018 Andrea Vázquez-Ingelmo, Francisco J. García-Peñalvo, Roberto Therón GRIAL Research Group. University of Salamanca andreavazquez@usal.es, fgarcia@usal.es, theron@usal.es
- 2. Outline • Introduction • Context • Materials and Methods • Results • Discussion • Conclusions
- 3. Introduction • Data is a vital resource for most activities, whether trivial or complex • Data supports informed decision making • It is necessary to exploit the collected data through its analysis in order to obtain valuable information • Information helps in the decision-making process, forming the basis for knowledge and wisdom generation
- 4. Introduction • Information dashboards play a key role in visualizing the outcomes derived from the data analysis through graphical resources • However, the final users of the dashboards can present different requirements from each other • It is important to consider all of them because the user experience depends on the achievement of its requirements
- 5. Introduction • Implementing customized dashboard for each user is not trivial and would consume significant development time • Information dashboards need to be flexible, maintainable and scalable in terms of content and design • Study fields like Software Product Lines (SPLs) and Model Driven Development (MDD) provide potentially useful solutions to produce dashboards with the previous characteristics
- 6. Introduction • The SPL approach aims to break down the solution in primary features that can be parameterized and composed to create variations adjusted to the target users in an efficient way • Software product lines foster an environment where reutilization and independent component configuration (through variability points) are the key features of its potential benefits
- 8. Context The Observatory for University Employment and Employability (OEEU) • An organization with the vision of becoming an information reference for understanding and exploiting knowledge about employment and employability • The Observatory collects and analyses data about graduates from Spanish universities to reach insights about these two fields • Different requirements regarding the presentation of results
- 9. Context • Users of the Observatory’s system: Spanish universities, administrators, general users, etc. • Necessity of a method to efficiently manage different requirements regarding the Observatory’s information dashboards • Application of the SPL paradigm on the Observatory’s system to automatically generate customized dashboards based on a set of requirements
- 10. Materials and Methods Base system • The Observatory counts on a base system developed with Django • The application of this paradigm is made on the presentation components that support the dashboards
- 11. Materials and Methods GraphQL API • GraphQL provides a language to query data from the backend • The development of a GraphQL API gives the chance to create flexible data flows between the dashboard components and the backend while maintaining high levels of decoupling
- 12. Materials and Methods Feature Model • It is necessary to study the commonalities and variability points of the different products of the SPL • The feature diagram allows to model the different products’ features
- 14. Materials and Methods Domain specific language (DSL) • A textual DSL was designed following the feature model • It allows the management of the requirements of the Observatory users regarding the dashboard elements in a centralized and high-level way • Implemented with XML technology • The DSL links the feature model with the source code generator
- 15. Materials and Methods Source code generator • Template-based code generator implemented with Python • The core assets of the product line are the different code templates that implement the dashboards’ components • Given a series of condition the functionalities are injected in the final dashboard’s source code based on the requirements specified
- 16. Materials and Methods Source code generator
- 17. Results • Only the specification of the dashboard configuration (through the DSL) is required to generate the whole set of source files needed • Any change on the requirements only involves the modification of the configuration XML files and the regeneration of the code • This approach increases • Efficiency • Maintainability • Scalability of the OEEU’s dashboards and its development
- 18. Examples of automatically generated dashboards
- 19. Results
- 20. Results
- 21. Results
- 22. OEEU
- 23. OEEU
- 24. OEEU
- 25. OEEU
- 26. OEEU
- 27. Discussion • The feature model increases the traceability between the formal specification of the SPL and its final implementation • The refactoring of the dashboard’s source code into templates and macros allows the reutilization of features and software components • Capacity of changing and introducing variations and new requirements without requiring significant development time • Improvement in terms of scalability and maintainability
- 28. Discussion • GraphQL has proved to be a nice option to connect the dashboard components to the backend, creating flexible data flows between them • The use of code templates to implement the core assets makes the solution applicable for other programming languages • Potential framework to perform A/B testing and to study which product configurations give better support to reach insights about a particular topic.
- 29. Conclusions • The SPL approach has been applied to address the OEEU issues regarding the customization and maintainability of their information dashboards • Future research lines might involve: • Automatic linkage between the feature model and the DSL to propagate any modification • Application of the obtained framework for A/B testing • Evaluations regarding human-computer interaction
- 30. Acknowledgements • The research leading to these results has received funding from “la Caixa” Foundation. • This work has been partially funded by the Spanish Government Ministry of Economy and Competitiveness throughout the DEFINES project (Ref. TIN2016-80172-R)
- 31. Application of domain engineering to generate customized information dashboards HCI 2018 Andrea Vázquez-Ingelmo, Francisco J. García-Peñalvo, Roberto Therón GRIAL Research Group. University of Salamanca andreavazquez@usal.es, fgarcia@usal.es, theron@usal.es