Tencent Yuanbao Gaokao Regional Passing Scores

Problem

Gaokao score-line questions are often hard to parse. Users may ask for “Shanghai, last three years, physics track, undergraduate batch” or omit the year, batch, and subject. The skill focuses on turning a natural-language query into structured lookup parameters before fetching results.

How It Works

• Place extraction: matches provinces against place.json, supports aliases and tolerant matching, such as turning “Minnan” into Fujian, and asks for the province when unclear.
• Year and subject track: resolves explicit or relative years, then infers the exam mode from place and year: traditional arts/science, 3+3, or 3+1+2. It maps terms like “science track” to physics or comprehensive where appropriate.
• Batch matching: reads luqupici_summary.json and matches admission batches; if only “undergraduate” is given, it expands to candidate batches such as general, arts, special programs, and sports.
• Query and answer: calls fetch_data.py, filters the returned data against the user's intent, and gives a concise answer.

Boundaries

It fits questions like “score line for a province, year, subject track, and batch.” If key parameters are missing, it may ask a follow-up question or omit the corresponding field. It is not a substitute for final admissions policy advice, detailed major comparison, or real-time regulatory interpretation.