MeetingNotes.MD

Meeting 20210115

Goal: learn from Sophie and Mikael about the data and papers they had collected before.

Feedback from the challenge owners

Sophie

• She provided a word file containing her tips on identifying the relevant literature (e.g. what keywords to look for)
• She used only Google Scholar and LITTERBASE for her search
• She focused only on the Atlantic
• She will share her final selection of papers. Unfortunately, she did not keep track of her search process and all irrelevant papers had been deleted
• In her experience, most data are available as supplementary materials online.

Mikael

• He will share information about the papers where he had to contact authors for data
• He will share the collected data set and the annotated data set
• He used mainly Google Scholar

Erik

• Would be nice to be able to flag papers where the data are not presented but might be available via contact with the authors
• But it's not so important
• Most important columns in Mikael's data spreadsheet: BCGHIJ (this might not be 100% correct; good to verify)

Next steps (discussion among the team)

• We will work together on labelling some data (esp. getting some irrelevant papers)
• Qixiang:
  • Prepare a data set with titles and abstracts, from Scopus
  • Check LITTERBASE
  • Work on getting meta-data for the papers
• Sebastian:
  • Prediction models
• Mehran:
  • Code for batch-downloading papers (both open-source and not)

Meeting 20210108

The presentation

Feedback from the challenge owners

Search terms

• important additional terms: nanoplastic, macro plastic
• less important: mesoplastic, contamination (too many items perhaps)
• leave out toxicants like PBDE

Database

• journals like marine pollution bulletin are important to be included

Pipeline

• add step 5: remove duplicate data

Labelled data

• should Sophie Schmiz on the data collection process and the final available data
• Mikael's paper points to 17 relevant papers

Other

• the expected time window of sampling is after 2008
• the expected number of relevant papers is in the hundreds (between 10 and 1000)
• the problem with LITTERBASE is that it is handcurated, but at some point we can contact them about using their findings as gold standard
• the ultimate goal is to automate as much as possible

Next steps

• Make an appoint with Sophie to learn about her data
• Identity those relevant papers and curate a data set (titles, abstracts and DOIs) - Qixiang
• Get a large collection of papers from databases (with export information like titles, abstracts and DOIs) - Sebastian
• Write a script to crawl meta data about the articles using the DOIs - Mehran & Qixiang
• Write a script or find a way to automate batch download of the full texts - Sebastian & Mehran
• Write a model to identify relevant papers - later
• See if it's possible to make use of ASReview - later

Kick-off Meeting 20201222

The ultimate goal (not of relevance to the current challenge)

• A 3d map of plastics in the ocean
• Eventually a virtual simulation of how plastics move
• Need as many observations of plastics as possible (to train the model)

Pre-existing work

• 2 master students going through literature manually
• The goal of the challenge is scraping of the literature
• There is a small database of the Mediterranean ocean now - curated data set ready by mid January

Main challenges

• How to automatically identify peer-reviewed papers that contain data on observations of plastic in the ocean and on beaches>
• How to automatically parse that data into a database?
• How to geo-tag the plastic observations to location and time of sampling?

Tips

• Start with challenge 1: identify relevant papers and if possible, automate extraction of the data sets
• Start with more recent papers: look for data availability statement
• We can get some expected meta data from Darshika.
• Darshika is first point of contact for any questions

To Do:

• Make a meeting on Jan. 8
• Ask for a couple of papers
• Start with Challenge 1
• Expect meta data afrom Darshika