Extracting science from the archive

Extracting science from the archive

• 1.
  Royal Society 2018, London,UK, 2018-06-12 Extracting Data from Early Scientific Diagrams Peter Murray-Rust1,2 [1]University of Cambridge [2]TheContentMine ContentMine extracts data from modern diagrams on a high-throughput scale. Do the same tools work for C19th scientific diagrams? Text, data on maps, chemical formulae, plots, phylogenetics, Images from ContentMine CC BY and Wikimedia CC BY-SA , ProcRoySoc PD pm286@cam.ac.uk peter@contentmine.org
• 2.
  Modern Diagram Mining 4500separate images Phylogenetic tree supertree A machine-compiled microbial supertree from figure-mining thousands of papers, Ross Mounce, Peter Murray- Rust, Matthew A Wills, 2017 https://riojournal.com/article/ 13589/
• 3.
  Original scanned bitmap Extractedby Tesseract Errors all due to untrained punctuation Text Mining bitmap
• 4.
  Original: V.D. vsTemperature Automatic Extraction: Grid, curve and points Plot Mining About 50% accurateProbably hand-drawn plot
• 5.
  binarization segmentation OCR Extraction ofchemistry: lines correct, some atom corruption
• 6.
  3 4 15 3 4 5 21 1 1 John’s Snow’s mapof deaths in Broad Street 1854 Onion-ring pixel analysis Segmentation and area/object count https://en.wikipedia.org/wiki/1854_Broad_Street_cholera_outbreak
• 7.
  https://en.wikipedia.org/wiki/Tree_of_life_(biology) Darwin’s Phylogenetic Tree Binarizednotebook segmentation Topological tree
• 8.
  https://commons.wikimedia.org/wiki/File:See_No_Evil,_Hear_No_Evil,_Speak_No_Evil.jpg Parodied byUser:petermr CC-BY SA 3 11 13 European Copyright Made Simple Mine no Content Link no Content Upload no Content Julia Reda’s explanation https://juliareda.eu/eu-copyright-reform/ and write to your MEP Peter Murray-Rust
• 9.
  All ContentMine Softwareis Free/Open • Contentmine.org • http://github.com/contentmine • http://github.com/petermr • http://discuss.contentmine.org/t/d3-text-processing- infrastructure/486/14 • Main site: Contentmine.org • Software: http://github.com/contentmine (production) and http://github.com/peterm (development forks) • Discussion and open notebooks: http://discuss.contentmine.org – http://discuss.contentmine.org/t/extracting-science-from-early-scientific- documents/613/ – http://discuss.contentmine.org/t/extracting-data-from-early-scientific- maps/614