Talk:Accuracy review

Jump to navigation Jump to search

Re: table of each word in article dumps indicating its age[edit]

Not an appropriate way to do this. See the ongoing m:Research:Content persistence. --Nemo 08:29, 13 February 2015 (UTC)

Do those methods accommodate reverted blanking and text passage moves? In any case, I've re-worked the description which was there when you responded, and am happy to cite [1] which seems fast enough and sufficiently robust against such distortions. It searches for context directly instead of building indexes keyed on context, which is probably faster. Thank you! Jsalsman (talk) 21:01, 13 February 2015 (UTC)

Future enhancement ideas[edit]

Here is a very rough outline of some ideas for future enhancement:

  1. measurement of editor workload and benefits
  2. new ways of identifying out-of-date, suspect, and confusing content
  3. stronger redundant anonymous voting systems to eliminate advocacy bias
  4. integration with Wikipedia Zero
  5. integration with educational freeware for computer-aided instruction
  6. paid reviewer experiments
  7. reviewer pre-qualification
  8. ongoing reviewer reputation tracking
  9. translation support
  10. double blinding with secret ballots and supervised exclusion from source pages
  11. tests on real-world controversial subjects
  12. independent subject matter expert evaluation

Jsalsman (talk) 19:04, 5 March 2015 (UTC)

Getting Started : Running the wikiwho code on your local PC[edit]

  1. Make sure you have a running version of Python 3 on your computer.
  2. Install mediawiki utilities by running the following command in your terminal:
    pip install mediawiki-utilities
    as outlined in this site
  3. Download the following two repositories and unzip hem into a single folder.
  4. Download the xml dump of any wiki article of your choice into the same folder. e.g.
  5. Run the file using the following command to get the authorship for all tokens having a particular revision id (686845449 in the train.xml case)
    python3 -i train.xml -o a -r 686845449
  6. You can use the following command to get the edit interactions of every revision to every other revision up to rev id 686845449:
    python3 -i train.xml -o r -r 686845449
  7. Further, create an account on PythonAnywhere ( and repeat the above steps in the bash console to get the same results.
  8. The next step would be to extract revision dates from the xml dumps.

— Preceding unsigned comment added by Prnkmp28 (talkcontribs)

Updates in the Phabricator task page[edit]

Added the tasks completed and the next tasks to be worked on. Also added the project repo and blog links.

— Preceding unsigned comment added by [[User:{{{1}}}|{{{1}}}]] ([[User talk:{{{1}}}|talk]] • [[Special:Contributions/{{{1}}}|contribs]])


BTW, this proposal will follow the same process as everyone else. The process does not include a speak now or never clause. Nemo 23:26, 21 February 2016 (UTC)