Releases: harmonydata/harmony
v1.0.7
What's Changed
- Add topics to the clustering by @tongyu0924 in #98
- chore: add fpdf2 to requirements and created PDF report export code by @dj0409 in #109
- feat: add PDF export functionality and test by @dj0409 in #110
- Issue #99: the package can't be installed and imported correctly in colab by @brandonHaipas in #106
- Replace the PDF parsing code with LLM model prediction by @montygole in #108
- fix PDF parsing - HuggingFace parser was crashing by @woodthom2 in #112
- feat: Add enhanced PDF harmonisation reports with graphics by @narendrakumar9867 in #114
- Add User-Defined Topics Demo to Harmony Walkthrough Notebook - Issue #105Add User-Defined Topics Demo to Harmony Walkthrough Notebook - Issue #105Demo user topics issue 105 by @DHANUSHRAJA22 in #116
- fix: add missing normalise_text function to resolve ImportError and CI failurefix: add missing normalise_text function to fix ImportErrorfix: add m… by @DHANUSHRAJA22 in #118
- Fix: Handle empty/whitespace inputs in process_questions and add ligh… by @GitAd7 in #119
- feat: Add PDF export functionality to main API (resolves #53)feat: Add PDF export functionality to main API (resolves #53)Update _… by @DHANUSHRAJA22 in #115
- feat: Add comprehensive HTML file support with intelligent parsing and minimal dependenciesfeat: Add HTML file support for instrument loadingfeat: Add HTML file support for instrument loadingFeature/html file support by @DHANUSHRAJA22 in #117
- feat: add question visualization GUI by @KonstantinosKorovesis in #113
- #38: Ingest an instrument from a URL by @evelynnchen-cmu in #75
- #120 bug fix: fixed process_questions in matcher.py so that unit test does not fail by @AlexNikic in #121
- Added matplotlib to tox by @AlexNikic in #122
New Contributors
- @dj0409 made their first contribution in #109
- @brandonHaipas made their first contribution in #106
- @narendrakumar9867 made their first contribution in #114
- @DHANUSHRAJA22 made their first contribution in #116
- @GitAd7 made their first contribution in #119
- @KonstantinosKorovesis made their first contribution in #113
Full Changelog: v1.0.5...v1.0.7
Contributors
Assets 2
v1.0.5
Added user-defined topics (Thanks @AlexNikic)
The matcher now allows users to pre-define topics, which the matcher will now use to tag the questions where the word-to-topic cosine similarities of the questions are above a certain threshold.
PR: #104
Issue: #82
Contributors
Assets 2
v1.0.4
Alex @AlexNikic has implemented a similarity matrix for response options - he submitted it this week to the Python library and API #101. Thanks Alex!
So in addition to the similarity matrix for questions, you can also get a separate similarity matrix that tells you how much the response options match for each question.
This is still to be implemented in R and the web front end.
Contributors
Assets 2
v1.0.3
What is new in this release:
The tool performs clustering of question items both with k-means and by a deterministic algorithm (thanks Krithika!)
The Python library generates crosswalks
There's a measure of instrument to instrument similarity
This is all updated in the Walkthrough document
- We are allowing the user to turn on/off the negation behaviour
- The tool has batching added to cope with larger lists of instruments (thanks makrianast )
v1.0.1
Update dependencies and fix to non-breaking versions (HuggingFace, Numpy, Lxml) - some third party dependencies had been updated and were breaking the install.
Also, we have just had three fantastic pull requests in the Python repository:
- Batch code courtesy of @makrianast: Allow batching of items when sent to LLM by Anastasia original issue
- Strip prefixes code courtesy of @abdullahwaqar feat(utils): Add strip_prefixes function to remove common question prefix by Abdullah original issue
- Crosswalk table generation courtesy of @vkrithika25 Added crosswalk table + unit tests by Krithika original issue
Contributors
Assets 2
v1.0.0
Add match_instruments functionality so that Harmony web API will now return matching items from catalogues such as UKLLC, Mental Health Catalogue
Updated PDF extraction to use Scikit-Learn instead of spaCy
Upgraded to Pydantic V2 (thanks to @op-cls)
Add integration function create_instrument_from_list so that user can import to web UI from Python (documentation here: https://harmonydata.ac.uk/integrating-with-harmony/ ) :
from harmony import create_instrument_from_list, import_instrument_into_harmony_web
instrument = load_instrument_from_list(["Do you currently smoke or have you ever smoked?", "[Do you currently use] nicotine replacement therapy?"])
web_url = import_instrument_into_harmony_web(instrument)
print (web_url)
v0.5.2
Return topic strengths (harmonydata/app#13)
Return Python package version from API harmonydata/harmonyapi#3 harmonydata/app#12