Intelligent Bid Opportunity Screening and Recommendation Engine

The Problem: Precise Filtering of Bidding Information

Enterprises seeking bidding opportunities face challenges of massive data volume, varied expressions of key requirements, and the possibility that procurement intent has not yet been formally announced. Manually browsing numerous announcements is inefficient and prone to missing projects in pre-procurement stages like Procurement Intent or Under Construction. This skill transforms a vague business description into structured search parameters and, after fetching real data via API, provides a traceable recommendation list rather than simple keyword matching.

How the Skill Works

The core is a structured process of Query Construction - Data Fetching - Intelligent Filtering - Structured Output.

1. Business Profiling & Keyword Expansion: Extracts product/service names, target customers, regions, etc., from user input to generate five types of search terms (Direct Subject Terms, Synonym Procurement Terms, Application Scenario Terms, Demand Trigger Terms, Exclusion Terms). It constructs search keywords using | for “OR” logic.
2. Full Pagination Data Retrieval:
   • Calls /bid/searchProjectApi for bidding projects (default: last 30 days) and /bid/searchNZJProjectApi for projects under construction (default: last 90 days).
   • Strictly follows full pagination rules: Continuously requests pages by incrementing pageId until hasNext=false, merges all page data, and deduplicates. Only calls detail APIs to supplement information for projects in the final display.
3. Intelligent Scoring & Recommendation:
   • Scores filtered candidates across six dimensions: Direct Business Match (40pts), Opportunity Stage (20pts), Timeliness (15pts), Regional Fit (10pts), Budget Fit (10pts), Evidence Completeness (5pts). Normalization is applied based on applicable dimensions.
   • Assigns recommendation grades: A (80–100), B (60–79), C (Below 60). The grade summary table and the count of items excluded by basic filtering are reported separately.
4. Explainable Output: Clearly distinguishes the source of business profile data (User-Provided / Default Conditions / Inferred Analysis). Each recommended project lists matching evidence, risks, and suggested next steps. It also declares the data coverage status (e.g., “Full data retrieved” or “Full retrieval incomplete”).

Applicable Boundaries and Considerations

• Dependent on a Valid API Key: A working BBIAO_API_KEY is required to call the API service; otherwise, queries cannot execute.
• Dependent on Real API Responses: Recommendations are based on real-time data returned by the API. Scores, grades, and evidence originate from this data; the skill itself does not predict procurement outcomes.
• Not an “Automated” Lead Generation Tool: The output is a list of leads for manual verification. Users must verify qualification, download attachments, contact the procurement entity, and follow up. The skill cannot guarantee contact availability, budget authenticity, or deal probability.
• Guidance and Interaction: Employs progressive guidance. Only the core business is required; other conditions (e.g., region, budget) are optional. The skill automatically applies default values for missing conditions and explicitly notes them in the output.