Public Procurement Analytics System
Analyzes notices and contracts with automated item classification, supplier comparison, and search for target buyers. The system helps explore the competitive landscape and pick tenders to pursue.
The task
Build an analytical system for studying the public procurement market: from assessing demand and the competitive landscape to finding buyers and selecting tenders that match a supplier’s product range.
Data and domain logic
The system continuously collects notices, contracts, participant records and award results from Russia’s Unified Information System for Procurement (EIS). This data supports the analysis of buyers, suppliers, prices and procurement history.
Item matching is one of the central problems. The same item can be described in many ways, so keyword search is complemented by normalization, fuzzy matching, machine classification and language models.
Implementation
We combined data collection and processing, matching models and an analytical interface. The system lets the user move from market research to specific tenders and potential buyers. Ambiguous matches are flagged for review.
The database holds 9 million contracts and 7.6 million notices under the two Russian procurement laws, 44-FZ and 223-FZ.