LLM-Based Query Expansion with Gaussian Kernel Semantic Enhancement for Dense Retrieval
Bibliographic record
Abstract
In the field of Information Retrieval (IR), user-submitted keyword queries often fail to accurately represent users’ true search intent. With the rapid advancement of artificial intelligence, particularly in natural language processing (NLP), query expansion (QE) based on large language models (LLMs) has emerged as a key strategy for improving retrieval effectiveness. However, such methods often introduce query topic drift, which negatively impacts retrieval accuracy and efficiency. To address this issue, this study proposes an LLM-based QE framework that incorporates a Gaussian kernel-enhanced semantic space for dense retrieval. Specifically, the model first employs LLMs to expand the semantic dimensions of the initial query, generating multiple query representations. Then, by introducing a Gaussian kernel semantic space, it captures deep semantic relationships among these query vectors, refining their semantic distribution to better represent the original query’s intent. Finally, the ColBERTv2 model is utilized to retrieve documents based on the enhanced query representations, enabling precise relevance assessment and improving retrieval performance. To validate the effectiveness of the proposed approach, extensive empirical evaluations were conducted on the MS MARCO passage ranking dataset. The model was systematically assessed using key metrics, including MAP, NDCG@10, MRR@10, and Recall@1000. Experimental results demonstrate that the proposed method outperforms existing approaches across multiple metrics, significantly improving retrieval precision while effectively mitigating query drift, offering a novel approach for building efficient QE mechanisms.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".