Project Template A: “Aspect-Based Sentiment → Tourism Improvement Strategy”

Project Template B: “Sentiment-Driven Tourism Demand Forecasting”

Research Notes

Understanding the Challenge Requirements

From the challenge page: itcodefair.cdu.edu.au

What the challenge involves

• You’ll use publicly available customer review data (Google Maps, TripAdvisor, Instagram, travel blogs) about tourism in the Northern Territory (or somewhere analogous). itcodefair.cdu.edu.au
• The goal is to extract tourist feedback and sentiment, identify what people are saying (key themes, pain points, strengths), and then propose recommendations that governments or businesses can act on. itcodefair.cdu.edu.au
• Ideally, you should also predict the impact of your recommendations on tourism numbers and economic value. itcodefair.cdu.edu.au
• Allowed techniques: text processing, sentiment analysis, topic modeling, clustering, recommendation systems, predictive modeling, and visualization. itcodefair.cdu.edu.au
• Submissions must be in Python (scripts or Jupyter Notebooks). itcodefair.cdu.edu.au
• The judging criteria include: dataset quality, creativity/originality, technical sophistication, contextual relevance/practicality, ethics, and presentation. itcodefair.cdu.edu.au

So your project needs to combine textual analysis (reviews/feedback) with actionable, predictive insights in the context of tourism and economic development. That gives you a somewhat wide space, but also constraints (you can’t just do a generic sentiment analysis; you need to tie it back to tourism policy/business decisions, and ideally quantify effects).

What makes a “good” project here (guiding principles)

When selecting or designing a case-study-style project to follow or replicate, keep these in mind:

1. Clear domain alignment

   The project should be about tourist reviews, destination experiences, or hospitality services so that your work maps directly to the NT tourism context.

2. From raw text → insights → decisions

   It should not stop at sentiment classification. Instead, it should go further: extract themes/aspects, quantify impact, simulate or forecast outcomes, and propose actionable interventions.

3. Use multiple sources or fuse data

   If possible, combining review data with external variables (e.g. visitor numbers, regional economic data, weather, transport stats) strengthens your predictive power and relevance.

4. Explainability and usability

   Since judges may include government/business people, make sure your findings are interpretable. E.g. “Sentiment about cleanliness is strongly correlated with revisit intention; improving cleanliness might increase visitor numbers by X%.”