Papers › A Model Ensemble Approach with LLM for Chinese Text Classification
A Model Ensemble Approach with LLM for Chinese Text Classification
Chengyan Wu, Wenlong Fang, Feipeng Dai, Hailong Yin
Automatic medical text categorization can assist doctors in efficiently managing patient information. By categorizing textual information such as patients’ descriptions of symptoms, doctors can easily find key information, accelerate the diagnostic process, provide superior medical advice, and successfully promote smart diagnosis and medical automated QA services. In this paper, an approach to medical text categorization is presented in the open-share task of the 9th China Conference on Health Information Processing (CHIP 2023), where complex textual relations are the two main challenges of this task. A model integration approach is proposed for this task, which can effectively solve medical text categorization through the complementary relationship of three different submodels. In addition, the solution provides external tools for targeted data enhancement for difficult samples that are hard to classify to reduce misclassification. Final results are obtained by the models through a voting mechanism. Experimental results show that the proposed method can achieve 92% accuracy and also prove the effectiveness of the model.
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Few-shot NER | CHIP-2023 | Qwen-7b-Chat | 1:1 Accuracy | 92 | #1 of 1 | Archive leaderboard | report |
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections