Building a biomedical tokenizer using the token lattice design pattern and the adapted Viterbi algorithm
Bibliographic record
Abstract
BACKGROUND: Tokenization is an important component of language processing yet there is no widely accepted tokenization method for English texts, including biomedical texts. Other than rule based techniques, tokenization in the biomedical domain has been regarded as a classification task. Biomedical classifier-based tokenizers either split or join textual objects through classification to form tokens. The idiosyncratic nature of each biomedical tokenizer's output complicates adoption and reuse. Furthermore, biomedical tokenizers generally lack guidance on how to apply an existing tokenizer to a new domain (subdomain). We identify and complete a novel tokenizer design pattern and suggest a systematic approach to tokenizer creation. We implement a tokenizer based on our design pattern that combines regular expressions and machine learning. Our machine learning approach differs from the previous split-join classification approaches. We evaluate our approach against three other tokenizers on the task of tokenizing biomedical text. RESULTS: Medpost and our adapted Viterbi tokenizer performed best with a 92.9% and 92.4% accuracy respectively. CONCLUSIONS: Our evaluation of our design pattern and guidelines supports our claim that the design pattern and guidelines are a viable approach to tokenizer construction (producing tokenizers matching leading custom-built tokenizers in a particular domain). Our evaluation also demonstrates that ambiguous tokenizations can be disambiguated through POS tagging. In doing so, POS tag sequences and training data have a significant impact on proper text tokenization.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".