Feedback from 2015-12-10-lifesciences workshop

[jcmolloy]

See repo and pad

General software functionality

• Lots of people wanted to use with existing PDF collections - this could be a specific module/tutorial. Need @petermr input to ensure people know how to try but are aware results could be patchy.
• Just a proposal: all the tools can be integrated into a tool sharing platform like 'Docker'. That way users would not have to run a virtual machine or install individual tools.
• Not sure if already noted: Powering off VM (e.g. after lunch) or opening editor ('Geany'), will reset keyboard input to non-US

  QS/Getpapers

• Lots of Qs about APIs/scrapers and links are not obvious without reading through all the software tutorial text. Could have a 'related links' at the top of the repo as well as the bottom or 'Are you looking for?' links.
• The VMs did not have scraper definitions so need to include these or give a clear instruction to download them with the following command: git clone https://github.com/ContentMine/journal-scrapers.git
• Trainer note: Need to be clear that while they're welcome to try their own material, they should follow all of the tutorials in order to use the papers for norma/AMI tutorials OR need bundles of files as input and output from each step in case it fails.
• Should flag in software tutorials that if they're using the vm they don't need to install the software and can use npm rather than sudo npm
• People wanted to work out how many papers they'd found before downloading them: getpapers --noexecute --query 'dinosaurs' etc
• If you get too many papers, try reducing the date range to a manageable number (see getpapers and EPMC query syntax). Our experience is ca 300 papers/min.
• To kill getpapers - Ctrl-C (maybe twice)

  Qs

• What is the .json file exactly (results)?
• Can we restrict the number of items to download?
• Can we use multiple search term Boolean queries to narrow results at getpapers stage?

  Norma

• Some people found that when running getpapers WITHOUT quickscrape, norma works but running quickscrape between getpapers and norma caused issues.

  Qs

• Where can we find a list for all the --transform arguments?

  AMI

• scholarly HTML link broken on ami tutorial, Input data section: https://github.com/ContentMine/workshop-resources/tree/master/software-tutorials/ami
• Find XML results files with data, group by ami command: ls -lht /////.xml
• Lots of participants very quickly wanted to summarise results.xml - could include work by Matthew Thomas after Wellcome workshop (link).

RegEx tutotial: Could you please give example results? I can run up to second last step and get regex and results folders, but all I opened manually are empty. How many results should there be for food?

Documentation

• Some specific comments that formatting all commands as code would be useful (I don't know how much code is not formatted as such, it might only have been a few lines)
• One participant commented 'Generally got a bit lost navigating between materials on GitHub pages'
• Overall feedback (personal, subjective): I got lost often, because I do not have a conceptual map of which step takes what input and creates what output after very simple steps (getpapers, scrape) were done. Vaguely getting there, but pace too fast. I was also often uncertain which directory I need to be in and how to access output. maybe provide hard-copy of pipeline for steps in workshop and vague 'directory' map of where stuff ends up (if the user records steps/doesn't change locations).
• One issue: If tutorial was carried out on own search (i.e. not on dinosaurs, you can no longer carry out later tutorials, e.g. I run norma on different topic, now I cannot follow ami2-species as I have no species in my text. Could you provide intermediary output folders, so that if you have to skip forward in tutorials, you can still follow (in future courses)? I have scholarly.html, but that doesn't contain species info.
• Overall, instructions are not very clear to me. Some command line commands are in `picture-type format (e.g. in a tree example, others in plain text, they are sequential, so you cannot follow later stages if you get stuck anywhere (not so great).