The repository contains files for extracting crystallization solvent from supplementary information PDFs using Natural Lanugage Processing (NLP). Check out the published article here.
The example code can be run on google colaboratory using this link
No local installation is required.
The crystallization solvent is extracted using rule-based method. The general rule is to search for solvent that follows the word crystallize or recrystallize. Shown below are the rules from solvent_parser.py file.
solvent = (solvent_name + Optional(R('/|and|:|-|&|') + solvent_name))(u'solvent').add_action(join)
prefix = (R('^(re)?(-)?crystalli[sz](ation|ed)$', re.I) +Optional(T('RB')) + Optional(T('IN')) + Optional(I('a') + Optional(T('CD') + W(":") + T('CD')) + (I('mixture') | I('solution')) + I('of')) + Optional(T('JJ'))+ Optional(R('\[|\(|\{')))
crys_sol = (prefix + solvent)(u'crys_sol')
Here solvent accounts for variations in mentions of crystallization solvent. The following scenarios are accounted
- single solvent (ex. methanol)
- multiple solvents (ex. methanol/hexane, methanol and hexane, methanol:hexane, methanol & hexane)
prefix captures the occurences of word (re)crystallize and the words that follow it. Examples
- recrystallized from
- crystallized from
- crystallized from mixture of
- crystallized (solvent)
- recrystallized in
crys_sol combines the prefix and solvent so that the text can be parsed to extract the solvent.
- solvent_nlp.py contains the code for parsing crystallization solvent
- webscraping.py contains the code for downloading SI in PDF format. See example.ipynb for usage
- extracted_data.csv has the data extracted from randomly selected 100 SI PDF files
- results.xlsx show the computation of F1 scores for the extracted data
- example.ipynb has the sample code for extracting the crystallization solvent from a PDF file.
Web scraping articles from publishers website may need prior approval. The web scraping code provide here by the authors is for demonstration purpose only.
If you find this useful, please cite the following article -
Smith, A.; Bhat, V.; Ai, Q.; Risko, C. Challenges in Information-Mining the Materials Literature: A Case Study and Perspective. Chemistry of Materials 2022, 34 (11), 4821–4827. https://doi.org/10.1021/acs.chemmater.2c00445.