Raw Data for Questionnaire Quality Evaluation Using LLM Fine-Tuning

169 professional psychological questionnaires used to fine-tune large language models for scale development. The dataset, created by Zhitao Yuan and last updated in April 2026, supports research into automating psychological assessment tools. It is a small 11.7 KB XLSX file containing raw data for evaluating model performance on text generation, scientific rigor, and cultural adaptability.

Use Cases

• Benchmarking LLM text generation quality for psychological questionnaires based on reported BLEU-4 and ROUGE-L metrics
• Evaluating the scientific rigor and logical consistency of AI-generated questionnaire items based on the described 28.6% improvement
• Assessing cultural adaptability in cross-regional expression conversion based on the reported over 85% accuracy
• Fine-tuning language models for specialized psychological assessment tasks based on the described instruction-tuning and human feedback reinforcement scheme

Strengths

• Includes a corpus of 169 professional psychological questionnaires for model training
• Model improvements are quantified with specific metrics: BLEU-4 increased by 0.05, ROUGE-L by 0.057
• Reports a 28.6% improvement in logical consistency and over 85% accuracy in cross-regional expression conversion

Limitations

• Row count is unknown, which may limit suitability assessment
• Column-level documentation is absent; field semantics must be inferred after download
• Description metadata is limited; actual data quality requires manual inspection after download

Provenance

Source

figshare, author Zhitao Yuan

Collection Method

Likely compiled from professional psychological questionnaires for a research study on LLM fine-tuning.

Time Range

Publication date suggests data was current as of April 2026.

Freshness

Last updated 2026-04-24 17:29:19; freshness should be verified

Geography

Cross-regional expression conversion mentioned suggests potential international scope, but specific geography is unknown.