Tencent Yuanbao Gaokao Score-to-Rank Lookup

Problem Being Solved

Gaokao score-to-rank questions are often phrased casually and mixed together. A user may ask “Where does 580 rank in Shanghai?” or “What was the science-track rank range in Shenzhen over the last three years?” These queries need natural-language parsing first, followed by careful handling of province, year, and subject-track policy differences.

How It Works

The skill focuses on five structured fields: place, year, classify, score, and rank.

• Place extraction: Supports abbreviations, nicknames, and city-to-province mapping, for example “Mogudu” to Shanghai and “Shenzhen” to Guangdong; it asks for clarification when the province is unclear.
• Year normalization: Converts expressions such as “this year,” “last year,” and “the last three years” into four-digit years, with fuzzy year expressions handled according to the defined rules.
• Subject-track conversion: Uses place and year to distinguish traditional arts/science, 3+3, or 3+1+2 modes, mapping user terms like “arts” or “science” to comprehensive, physics, or history where appropriate.
• Score and rank handling: Only numeric values are accepted, and Chinese numeric expressions are converted to integers, such as “50k” to 50000.
• Data retrieval: After the fields are confirmed, it calls ./fetch_data.py to query data and filters results according to the intended subject track.

Scope and Caveats

This skill fits structured queries such as “score to rank,” “rank to score,” or “rank range for a year and track.” It asks for the missing province when needed, and it asks for score or rank if the user does not provide one. Data access depends on the scripted workflow, so bypassing the script to fetch full datasets directly is not recommended.