UdA Connect: XAI and Machine Learning at the service of Research
INGENIUM is an alliance of ten institutions of higher education from ten European countries, including the University G. D'Annunzio of Chieti-Pescara, which is responsible for the promotion project of the alliance itself, "BI4E: Boosting INGENIUM for Alliance”, funded by the European Community.
Client
INGENIUM - Università degli Studi “G.D’Annunzio”
Industry
Research & Education
Year
2023
The initial situation
The initial scenario is one that arises upon the issuance of a new research grant, when there is a need to create a new team or integrate new roles into an existing team with the required expertise. Those responsible for this task find themselves navigating through different tools - LinkedIn, Scopus (a research publication portal), Orcid (researcher registry portal) - to conduct the initial research phase. Once the relevant profiles are retrieved, it is necessary to contact various professionals and verify their availability. The process concludes with the formation of the team, after several offline touchpoints and online portals, with an additional time commitment, not always available during the grant application phase.
20tab worked on a platform that could facilitate the connection among all professors and researchers active in the 10 universities belonging to the INGENIUM consortium, promoting collaboration on research projects. The goal was to form more relevant multidisciplinary teams aligned with the research topics addressed.
Main technical activities
Development
Discovery
Tools and practices
Design Thinking
DevOps
Opportunity Solution Tree
Customer Journey Maps
FURPS+ Model
User Flow
Lean Canvas
User Story Mapping
Implemented activities
A system that increasingly approximates the human search process has been implemented to achieve precision in recommended search results. Models of Machine Learning and Explainable Artificial Intelligence (XAI) processes were chosen to make the solutions and choices provided by the models transparent.
Let's go through all the steps of the process.
The Discovery and assessment phase allowed us to begin understanding the characteristics that the "UdA Connect" platform should have. Stakeholders involved were identified through a Stakeholder Map, and the project scope, along with its problems and solutions, was defined. We discussed input/output and functionalities, outlining the envisioned functioning of the portal.
Subsequently, an in-depth analysis of the context was conducted, identifying direct and indirect competitors. Through benchmarking, various aspects were analyzed, including user interface, accessibility, and search input and output. Solutions with similar functionalities in different sectors were also examined.
This allowed us to establish a user research plan, hypothesizing proto-personas and conducting interviews to trace the User Journey As-Is and validate the personas themselves.
Co-creation sessions, along with information collected from users, delved into the portal's functioning, highlighting potential areas for improvement and communication methods for explainability.
After reviewing the User Journey and organizing the Opportunity Solution Tree, the team worked on the User Flow to investigate and resolve any navigation friction points.
Subsequently, high-fidelity Wireframes were created, streamlining the actual implementation.
This work led to a product Roadmap, divided into 3 releases, and the continuation of the actual platform development.
From a technical standpoint, the team focused mainly on XAI model development and web development. Later, there was integration of the XAI module into the web application, the data retrieval and management phase, and finally, deployment into the production environment.
Implemented Technologies
Final result
The 20tab team has delved into software development in the field of eXplainable Artificial Intelligence (XAI), releasing a platform that allows users to input discourse in the research field and obtain output relevant to the initial research intent.
By entering keywords related to the research objective, the system provides a list of all researchers related to the topic. It is then possible to delve into each researcher's expertise by viewing their studies and publications. At this point, the portal also allows users to access contact information, providing links to external platforms.
"Uda Connect" processes this data by integrating the ORCID code for each profile and uses information from researchers at the 10 European Universities belonging to the INGENIUM alliance as sources. The data loading means were provided by 20tab, and the data source is Scopus.
This type of development has streamlined and expedited the University's process in the personalized search for members of research groups aligned with the goals of the research.