Extending TriRAG for Advancing Retrieval-Augmented Generation Method with Triple-Based Knowledge Graphs for Improved Question Answering
Bibliographic record
Abstract
While large language models exhibit broad capabilities, they often require supplementation for optimal performance on domain-specific tasks. To address this, we built TriRAG, an innovative enhancement to traditional Retrieval-Augmented Generation. This paper presents an extended version of our earlier TriRAG work. TriRAG integrates a structured knowledge graph composed of semantic triples derived from text, significantly improving the performance of LLMs on multiple-choice question-answering tasks. Our method dynamically converts relevant text into triples, embeds them into vectors, and retrieves the most useful triples for a given question by calculating vector similarities. By having the triple-based approach instead of the conventional text-based retrieval approach, TriRAG enables more precise and efficient information retrieval. This directly enhances the accuracy of LLM-generated responses in multiple-choice question-answering tasks. We evaluate TriRAG using the Textbook Question Answering dataset, demonstrating consistent improvements over traditional RAG methods across leading LLMs, including Gemma, Llama, and ChatGPT variants. Experimental results and ablation studies confirm that our triple-based system enhances both retrieval accuracy and processing efficiency, leading to better overall model performance.
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 | 0.002 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 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".