Section A: Project Information
This project focuses on the implementation of an Article Hub on a Data Visualisation portal. It is a dedicated publication hosting service that hosts a repository of data visualisation publications, providing users with a rich collection of publications to learn about data visualisation techniques and data storytelling. One core feature implemented is the automated feedback generation system which uses Large Language Models (LLMs) to analyse and evaluate publications uploaded. The system encourages continuous and iterative improvements by allowing the user to edit and reupload their publication.
The system uses a React frontend styled with Material UI for their ability to create a dynamic and interactive web application. FastAPI is used to manage backend processes, including asynchronous feedback generation handled via FastAPI’s BackgroundTask. Amazon S3 is used to store uploaded publication files reliably and at scale, while user-generated content is managed via PostgreSQL. Authentication and authorization are handled through Keycloak, ensuring secure user access.
The platform aims to make learning about Data Visualisation and the creation of such articles accessible. With the automated feedback generation system, the Article Hub provides users with almost instant feedback. This reduces the reliance on human evaluators, while promoting high-quality publication standards. It has the potential to scale into a comprehensive, community-driven repository for learning, sharing, and improving data visualisations publications.
Section B: Participant Information
| Title | First Name | Last Name | Organisation/Institution | Faculty/Department/Unit | Phone Number | Contact Person / Team Leader | |
|---|---|---|---|---|---|---|---|
| Ms. | Xin Yi, Nicole | Lim | National University of Singapore | School of Computing | e0773160@u.nus.edu | +6591128301 |
|
| Dr. | Bimlesh | Wadhwa | National University of Singapore | School of Computing | bimlesh@nus.edu.sg | +6565162973 |
Section C: Project Details
The ability to create data visualisation and to communicate data-driven insights is gaining prominence in recent years – the demand for research analysts is expected to grow 25% between 2020 and 2030, and research analysts are increasingly expected to possess data storytelling skills. However, acquiring such a skill set remains challenging, especially for new learners, due to the lack of structured guidance and accessible, real-world examples.
By including automated feedback and allowing users to view publications published by others, this project creates an environment conducive for self-directed learning. The automated feedback generation system provides users with timely and quality feedback, which allows users to continuously and iteratively improve the overall quality of their publication. As such, learners will be better supported in their efforts to improve their communication of data insights.
Moreover, the platform provides learners with a repository of publications that illustrate diverse ways of visualising data and communicating data-driven insights effectively. This exposure helps bridge the gap between theoretical knowledge and practical application by allowing users to see real-world applications of concepts they are learning or have learnt.
This project aims to help learners develop the necessary skills to produce compelling, data-driven narratives. It supports the hypothesis that example-driven and feedback-oriented learning will lead to improved proficiency in data storytelling.
FastAPI serves as the backend development framework, which allows for the development of the required APIs and the usage of FastAPI’s BackgroundTask to generate the feedback in the background (asynchronously). This project leverage on LLMs and relevant prompt engineering patterns – persona, template, and reflection patterns – to generate relevant feedback for the publications. The frontend uses React with Material-UI (MUI) components to create a dynamic and interactive web application.
To validate the market demand for an Article Hub that leverages on LLMs for content evaluation, several approaches can be adopted. First, we can conduct user surveys to gauge interest in LLM-generated feedback and its perceived usefulness in enhancing data storytelling skills. We can also conduct pilot runs with educational institutions to observe how such tools impact student learning outcomes, particularly in courses focusing on data visualisation.
Core functionalities include:
- A user profile page for users to manage their publication(s)
- A form for users to upload their publication onto the platform
- An automated feedback generation system that evaluates a publication based on a set of metrics – relevance, language, consistency, and coherence
- A publishing process which considers the feedback given to a publication to determine whether the publication can be published on the platform.
To assess the usability of the platform, we can perform user testing to evaluate the web application based on a set of metrics:
- The ease of use of the platform
- The effectiveness of the LLM-generated feedback in helping users improve their publications
NA
This project introduces an innovative approach in supporting the learning and development of data storytelling skills by leveraging on LLMs to automate the evaluation of publications uploaded by the user. Traditionally, human reviewers / evaluators are required when assessing the quality of written content, which is often time consuming and difficult to obtain. By incorporating LLMs into the feedback generation process, the system delivers almost instant, tailored feedback to users based on specific evaluation criteria. This allows users to learn what was done well and what can be done better immediately after submission, allowing them to revise and refine their work iteratively.
To promote scalability, the backend is containerised using Docker, which simplifies the deployment and allows for horizontal scaling when user demand increases. Containerisation also allows for consistent behaviour across different platforms. Additionally, we have used Amazon S3 for file storage, ensuring fast and reliable access to uploaded content while supporting scalability in handling large volumes of data.
To ensure long-term sustainability, the project uses the simplest LLM possible to generate the feedback. Hence, less resources and computing power is required to generate the feedback for the publications. This reduces the overall carbon footprint associated with running LLM-based applications.
This project aims to democratise the access to feedback and resources in the field of data visualisation and data storytelling. This allows users, even those without institutional support, to learn and ultimately be proficient in communicating data-driven insights.
The social impact can be measured through:
- User satisfaction: Analyse how useful the user finds the automated feedback and whether they feel more confident in practicing their ability to communicate data-driven insights to various stakeholders
- Learning outcomes: Analyse whether the user feels that the feedback generated truly help them to improve their publication
To ensure that the platform continues to evolve in respond to the changing user needs, we need to incorporate user feedback forms. The feedback collected will guide future enhancements made to the platform in response the evolving user needs and challenges.
Personal Information Collection Statement (PICS):
1. The personal data collected in this form will be used for activity-organizing, record keeping and reporting only. The collected personal data will be purged within 6 years after the event.
2. Please note that it is obligatory to provide the personal data required.
3. Your personal data collected will be kept by the LTTC and will not be transferred to outside parties.
4. You have the right to request access to and correction of information held by us about you. If you wish to access or correct your personal data, please contact our staff at lttc@eduhk.hk.
5. The University’s Privacy Policy Statement can be access at https://www.eduhk.hk/en/privacy-policy.
- I have read and agree to the competition rules and privacy policy.