This is a research project to establish a digital marketplace platform that enables Buyers and Suppliers to connect through an online portal.
Develop a recommendation algorithm for a digital marketplace platform that connects New Zealand public sector agencies (buyers) with various suppliers (sellers) by providing personalized and relevant product or service suggestions. The algorithm should effectively incorporate Future Procurement Opportunities https://www.gets.govt.nz/FutureProcurementOpportunitiesIndex.htm to help suppliers better prepare for and align with upcoming demands from public sector agencies.
- Take into account users' preferences, behavior, and past interactions on the platform, considering both buyers' and sellers' perspectives.
- Utilize item similarity, collaborative filtering, and popularity to generate recommendations that cater to the specific needs of public sector agencies.
- Leverage Future Procurement Opportunities to assist suppliers in anticipating upcoming demands and adjusting their offerings accordingly.
- Consider seasonality, trending items, and inventory levels when recommending products or services.
- Balance the recommendations with a mix of in-network (suppliers the agency has previously worked with) and out-of-network sources (new or undiscovered suppliers) to promote diversity and discoverability.
- Continuously learn and adapt to user feedback and engagement metrics to optimize recommendations, ensuring that the platform remains up-to-date and responsive to the needs of both public sector agencies and suppliers. With this refined problem statement, we can better focus on building a recommendation algorithm tailored to the specific needs of your digital marketplace platform, connecting New Zealand public sector agencies with suitable suppliers.
for the digital marketplace platform that connects buyers and sellers. We can assign weightings to these attributes to create a scoring system that helps generate personalized and relevant product/service suggestions.
attributes and their respective weightings for the project:
- Interaction between buyer and supplier (0.75x) -Direct communication between the buyer and supplier, such as messaging or inquiries about specific products/services.
- Purchase history (0.135x) - Previous transactions between the buyer and supplier, indicating a successful business relationship.
- Buyer's time spent on the supplier's profile or product/service page (>2 mins: 0.10x) - An indicator of the buyer's interest in the supplier's offerings.
- Inclusion of a supplier in the buyer's favorites or shortlist (0.01x) - Shows buyer's intent to keep a supplier in consideration for future procurement opportunities.
- Buyer's rating or review of a supplier (0.005x) - Reflects the buyer's satisfaction with a supplier's products/services.
- Item similarity (0.03x) - The degree to which the supplier's products/services match the buyer's preferences and past purchases. This could be measured using text analysis or categorical similarity.
- Supplier's response time (0.02x) - The average time it takes for a supplier to respond to a buyer's inquiry or request. A faster response time could indicate better customer service and communication.
- Supplier diversity (0.015x) - A measure of whether the supplier is underrepresented (e.g., Māori and Pasifika suppliers). This can help agencies meet their diversity goals and support social procurement outcomes.
These weightings can be used to score and rank suppliers based on their relevance to a specific buyer. By incorporating these attributes and their respective weightings into the scoring system, the recommendation algorithm becomes more comprehensive and better suited to the project's goals. Ultimately, the algorithm will generate more accurate and relevant product/service suggestions for New Zealand public sector agencies, connecting them with appropriate suppliers on the digital marketplace platform.