Unlocking Web Histories: Leveraging LLMs and RAG to Transform Discovery in Web Archives
Bibliographic record
Abstract
This paper explores how Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) can be applied to improve access to web archives. These collections are often difficult to navigate due to their complexity and the limitations of traditional search tools. The author examines how RAG can help address concerns around trust and transparency in AI by grounding LLM outputs in external, curated sources. At the University of Victoria Libraries, a custom RAG pipeline was developed to build on tools like WARC-GPT. This pipeline enables natural language querying with source attribution and incorporates optimizations in data preprocessing, chunking strategies, and hardware acceleration. When tested on real-world web archives of cultural significance, it demonstrated improved retrieval accuracy and computational efficiency. The discussion also considers the broader implications for digital preservation in an age of uncertainty, emphasizing the importance of trust, ethics, sustainability, and continued human oversight in AI-powered discovery. The findings suggest that RAG offers a promising way to unlock the value of underused digital heritage collections while upholding the foundational values of research libraries, including long-term preservation, equitable access, and responsible stewardship of historic materials.
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 | 0.005 | 0.024 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.005 | 0.010 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
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".