Interactive Campus Crime Data Analytics: A Hybrid LLM-RAG System with Query Routing
Bibliographic record
Abstract
Ensuring campus safety remains a pressing concern for universities, as it is a foundational requirement for student well-being, yet most existing crime reporting systems fail to provide timely, interactive, or actionable insights, leaving critical gaps in awareness and response. This paper introduces a Retrieval-Augmented Generation (RAG) system for interactive campus crime analysis, tailored to support real-time student and staff queries. Our system utilizes university incident reports from York University and enables natural language queries by combining semantic search over preprocessed incident summaries with large language model (LLM) generation. To address RAG's limitations in structured factual queries, we introduce a hybrid question and answer (QA) router that distinguishes between generative and structured queries, routing the latter to rule-based logic. We also implement and test prompt engineering techniques to reduce LLM hallucinations. We qualitatively benchmark open-source embeddings and LLM model combinations and assess their performance on real-world queries. Among the evaluated LLMs are Gemma3:1B and Deepseek-r1:1.5B, which have delivered reliable and contextually grounded responses across narrative queries. This work demonstrates a novel and practical approach to crime awareness and decision support in campus environments.
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 | 0.003 | 0.010 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.003 | 0.005 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.009 | 0.004 |
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".