Bibliographic record
Abstract
Word embedding has been a significant breakthrough in natural language processing (NLP). Although word representation has improved remarkably and resulted in better performance in downstream NLP applications, interpretability of word embeddings remains a challenge. Post-processing techniques have been developed to improve the quality of word embeddings by fine-tuning real-valued vectors and eliminating artifacts that arise due to flaws in the training corpus. In this thesis, we propose a simple method to improve the quality of word embeddings and reveal their hidden structure using a post-processing technique. We deploy co-clustering techniques to detect sub-matrices between word meaning and specific dimensions. Furthermore, pre-trained word embeddings based on neural networks often suffer from representation degeneration issues, especially when trained on large datasets. We propose a novel method Orthogonal Auto Encoder with Variational Dropout (OAEVD), which utilizes orthogonal autoencoders and variational dropout techniques to enhance word embedding. The orthogonality constraint encourages more diversity in the latent space, and variational dropout makes the embedding more robust to overfitting. Empirical evaluation on a range of downstream NLP tasks shows that our proposed method effectively improves the quality of pre-trained word embeddings and pre-trained language models. In addition, We introduce a technique to enhance the accuracy of word embeddings by incorporating external knowledge in the form of WordNet pairs to improve the accuracy of word embedding on several NLP tasks.
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 | 0.009 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.010 | 0.007 |
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".