Enhancing the Reliability of Large Language Models in Specialized Domains by Balancing Internal and External Knowledge
Bibliographic record
Abstract
With the widespread application of large language models, numerous approaches have been proposed to enhance their performance. Most existing works focus on general-purpose tasks. Although some domain-specific approaches have emerged, they typically rely on enriching LLMs’ knowledge through additional training or retrieving external information, without effectively leveraging internal knowledge or balancing it with external sources. To address this, we propose a framework that operates effectively with only a few representative training samples (no more than one order of magnitude) to adapt to specialized tasks. This method is well-suited for scenarios with limited data, where fine-tuning is infeasible. Our method first adopts LLM-generated few-shot examples from training set, involving Chain-of-Thought, to maximize the utilization of internal knowledge. Then, uncertainty-guided retrieval-augmented generation is employed to selectively incorporate external information. This method balances internal and external knowledge, thereby reducing hallucinations arising from knowledge gap and errors caused by inaccurate retrieved context. Our method is compared with several representative, general-purpose methods on domain-specific datasets (law and finance). Experimental results show that our approach outperforms existing training-free, general-purpose methods across most tasks.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.006 | 0.027 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.004 | 0.003 |
| Research integrity | 0.003 | 0.004 |
| Insufficient payload (model declined to judge) | 0.002 | 0.003 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".