Interactive Tkinter Data visualizer for browsing/annotation for Instagram Vaccination Misinformation project. Requests information from Google cloud storage, stores updated Annotation responses.
Download all the files. Extract the folder. Execute bash script mac-setup.command by clicking icon and chosing to run in Terminal. Alternatively open a Terminal, change to application directory, ex: type cd /home/<>/Downloads/Data_viewer (or your folder path) && ./mac-setup.command
Install Python 3.7.3 here https://www.python.org/ftp/python/3.7.3/python-3.7.3-macosx10.6.pkg (64/32 bit installer for Mac OSX) or https://www.python.org/ftp/python/3.7.3/python-3.7.3-macosx10.9.pkg (64 bit installer OSX). Run the Installer.
Running internet connection required through out. API launch takes some time. Annotator launches with initial window to ask user details to keep track of annotators and jobs. Do not leave blank. Choose unique username to record responses Dashboard has options to browse (next to browse), delete (irrelevant samples), and label polarity, category and misinformation content of posts. Textbox displays caption, image-text rendering and information extracted from URLs Counter to track number of posts labeled by every user. Delete images irrelevant to content, skip if not confident of appropriate label or content has a mismatch between actual image-text and rendered image-text/image object labels. Deleted images are stored separately and can be retrieved as necessary. Choose between several categories of posts as you may deem relevant. Multiple choices if indicated, are recorded. Final responses are not recorded until you "submit". Shift between selections of polarity and content label if required before "submitting" final selections. Or use "reset category" to re-enter category choices before submission. Interface closes when annnotation of all posts is completed by a user or remaining posts are being viewed by other users. Message is displayed in Terminal accordingly. If runtime issues are encountered,open an "issue" thread (GitHub account required) or email repo owner.
10/18/2019 1.Tracking deletion of posts irrelevant to the scope of study (for example room cleaners from brand "Vax" or the band The Vaccines) 2. Updating log files upon termination. I added some signal handlers which should work for Linux and Mac OS. but it is recommended that users close the API interface before closing the terminal so that their activity gets refreshed for the next session. 3. Removing inclusion of URL data, something I discussed with Dr. Zhang, so that users can navigate between samples faster without having to download HTML contents.
11/25/2019: Style/Content/Topic options. Faster navigation. Completed posts count displayed is an estimate and may not increment evenly