Table 3.
Technique for order of preference by similarity to ideal solution (TOPSIS) composite performance scores and tier classifications across conditions.
| Condition | Model | TOPSIS score (95% CI) | Tier | Rank |
| Chinese_Expert | DeepSeek | 0.97 (0.82-1.00) | Tier 1 | 1 |
| Chinese_Expert | Qwen | 0.95 (0.84-0.98) | Tier 1 | 2 |
| Chinese_Expert | ChatGPT | 0.90 (0.79-0.97) | Tier 1 | 3 |
| Chinese_Expert | Grok | 0.75 (0.46-0.80) | Tier 2 | 4 |
| Chinese_Expert | Doubao | 0.70 (0.39-0.76) | Tier 2 | 5 |
| Chinese_Expert | Hunyuan | 0.70 (0.36-0.74) | Tier 2 | 6 |
| Chinese_Expert | Gemini | 0.65 (0.25-0.70) | Tier 2 | 7 |
| Chinese_Expert | Claude | 0.54 (0.03-0.58) | Tier 2 | 8 |
| Chinese_Expert | Kimi | 0.40 (0.00-0.43) | Tier 2 | 9 |
| Chinese_Expert | ERNIE X1 Turbo | 0.13 (0.00-0.32) | Tier 3 | 10 |
| Chinese_Expert | ChatGLM | 0.04 (0.00-0.35) | Tier 3 | 11 |
| Chinese_No-context | ChatGPT | 0.95 (0.88-1.00) | Tier 1 | 1 |
| Chinese_No-context | DeepSeek | 0.93 (0.82-1.00) | Tier 1 | 2 |
| Chinese_No-context | Qwen | 0.90 (0.81-0.98) | Tier 1 | 3 |
| Chinese_No-context | Doubao | 0.88 (0.77-0.93) | Tier 1 | 4 |
| Chinese_No-context | Gemini | 0.71 (0.24-0.78) | Tier 2 | 5 |
| Chinese_No-context | Claude | 0.70 (0.24-0.77) | Tier 2 | 6 |
| Chinese_No-context | Hunyuan | 0.67 (0.20-0.73) | Tier 2 | 7 |
| Chinese_No-context | Grok | 0.65 (0.09-0.71) | Tier 2 | 8 |
| Chinese_No-context | ERNIE X1 Turbo | 0.14 (0.02-0.37) | Tier 3 | 9 |
| Chinese_No-context | Kimi | 0.07 (0.00-0.38) | Tier 3 | 10 |
| Chinese_No-context | ChatGLM | 0.05 (0.00-0.39) | Tier 3 | 11 |
| English_Expert | ChatGPT | 0.93 (0.76-0.94) | Tier 1 | 1 |
| English_Expert | Qwen | 0.92 (0.79-1.00) | Tier 1 | 2 |
| English_Expert | DeepSeek | 0.86 (0.63-0.97) | Tier 1 | 3 |
| English_Expert | Doubao | 0.75 (0.36-0.83) | Tier 2 | 4 |
| English_Expert | Hunyuan | 0.74 (0.33-0.84) | Tier 2 | 5 |
| English_Expert | Claude | 0.70 (0.29-0.80) | Tier 2 | 6 |
| English_Expert | Gemini | 0.70 (0.25-0.78) | Tier 2 | 7 |
| English_Expert | Grok | 0.67 (0.17-0.75) | Tier 2 | 8 |
| English_Expert | Kimi | 0.64 (0.23-0.73) | Tier 2 | 9 |
| English_Expert | ERNIE X1 Turbo | 0.26 (0.00-0.35) | Tier 3 | 10 |
| English_Expert | ChatGLM | 0.00 (0.00-0.40) | Tier 3 | 11 |
| English_No-context | ChatGPT | 0.96 (0.88-1.00) | Tier 1 | 1 |
| English_No-context | Gemini | 0.90 (0.85-0.98) | Tier 1 | 2 |
| English_No-context | Qwen | 0.88 (0.83-0.98) | Tier 1 | 3 |
| English_No-context | Grok | 0.88 (0.81-0.96) | Tier 1 | 4 |
| English_No-context | Doubao | 0.85 (0.73-0.94) | Tier 1 | 5 |
| English_No-context | Hunyuan | 0.78 (0.49-0.87) | Tier 2 | 6 |
| English_No-context | Claude | 0.70 (0.29-0.78) | Tier 2 | 7 |
| English_No-context | DeepSeek | 0.70 (0.22-0.78) | Tier 2 | 8 |
| English_No-context | Kimi | 0.23 (0.02-0.37) | Tier 3 | 9 |
| English_No-context | ERNIE X1 Turbo | 0.22 (0.06-0.38) | Tier 3 | 10 |
| English_No-context | ChatGLM | 0.00 (0.00-0.40) | Tier 3 | 11 |