From Randy Miller ...

I have just submitted the final progress report for a recently completed NLM R01 grant. In it I declared the attached hand-edited set of 15,000 synonyms that I had created during the project as one of its inventions, and also declared that we will release it in the public domain.

I had shared an incomplete version of the synonym set with Clem, Olivier and the UMLS group there last year. This is now the completed version. It is derived from the original synonym set used in QMR, from various UMLS sources, from WordNet, and from my own clinical knowledge.

Sample contents are as follows:

abraded|scratch
abrasions|scratch
abrasion|scratch
abruptly|abrupt
abruptly|sudden
abrupt|sudden
abscess apical alveolar|abscess periapical
abscess apical dentoalveolar|abscess periapical
abscess apical|abscess periapical
abscess dentoalveolar|abscess periapical
abscess extradural|abscess epidural
abscesses apical alveolar|abscess periapical
abscesses apical dentoalveolar|abscess periapical
abscesses|abscess
abscess|abscesses
absence of bladder continence|incontinence urine
absence|absences
absence|absent
absent salivary secretion|xerostomia
absent|absence
absent|absences

Explanation: Strictly speaking, these are one-way mappings from the term on the left of the dividing bar ( | ) to the term on the right of the bar. In about 5-10% of instances, the mapping goes from a broader term to a narrower term that is not true in the converse direction. For the other 90-95% of mappings, they are reversible (i.e., valid in either direction).

A few notes about the content:

1) the terms on the right hand side of the mappings are limited to those matching content of the QMR lexicon. Otherwise this set of synonyms, unconstrained, would be several fold larger. It is nevertheless unique and generally useful in most other contexts in its present form.

2) if a mapping can be constructed word-by-word (or term plus term) from existing mappings in this set, then the “duplicative” mapping is in general excluded from the set. For example, if A, B, C, D, and E are words, and there are already mappings A B|D and C|E, then the mapping A B C|D E is not included in the set.

3) I personally (and painfully) reviewed all of the mappings for validity, and excluded mappings that make sense in common English but which have no relevance clinically. I may have missed a few things while bleary-eyed, but in general I think I caught most of the irrelevant mappings. Conversely, I added a number of mappings based on my own clinical knowledge that were not present in any of the other sources I used.

4) Please feel free to distribute this synonym set in any way that makes sense to NLM. I can write better documentation if you need it. Please also cite NLM grant R01 LM 010828 as the source of the funding that supported its development.

Thanks and best regards,

Randy