Utilizing Language Models to Expand Vision-Based Commonsense Knowledge Graphs
Bibliographic record
Abstract
The introduction and ever-growing size of the transformer deep-learning architecture have had a tremendous impact not only in the field of natural language processing but also in other fields. The transformer-based language models have contributed to a renewed interest in commonsense knowledge due to the abilities of deep learning models. Recent literature has focused on analyzing commonsense embedded within the pre-trained parameters of these models and embedding missing commonsense using knowledge graphs and fine-tuning. We base our current work on the empirically proven language understanding of very large transformer-based language models to expand a limited commonsense knowledge graph, initially generated only on visual data. The few-shot-prompted pre-trained language models can learn the context of an initial knowledge graph with less bias than language models fine-tuned on a large initial corpus. It is also shown that these models can offer new concepts that are added to the vision-based knowledge graph. This two-step approach of vision mining and language model prompts results in the auto-generation of a commonsense knowledge graph well equipped with physical commonsense, which is human commonsense gained by interacting with the physical world. To prompt the language models, we adapted the chain-of-thought method of prompting. To the best of our knowledge, it is a novel contribution to the domain of the generation of commonsense knowledge, which can result in a five-fold cost reduction compared to the state-of-the-art. Another contribution is assigning fuzzy linguistic terms to the generated triples. The process is end to end in the context of knowledge graphs. It means the triples are verbalized to natural language, and after being processed, the results are converted back to triples and added to the commonsense knowledge graph.
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 | 0.000 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
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".