Bridging Qualitative and Quantitative Methods for User Modeling: Tracing Cancer Patient Behavior in an Online Health Community
Repository for analysis code related to a Spring 2018 study investigating the health journeys of authors on CaringBridge.org.
Originally submitted to ICWSM in January 2019 and accepted July 2019 for presentation at ICWSM 2020 in June 2020.
As described in the paper, CaringBridge data used for analysis is not being released publically for ethical reasons.
For any questions or additional information, contact the corresponding author: [email protected]
Published paper (in ICWSM 2020 proceedings): https://z.umn.edu/bridging2020
Paper presentation (video): https://z.umn.edu/bridging2020video
Preprint: https://z.umn.edu/levonianIcwsm2020Preprint
Author website: https://z.umn.edu/zlevonian
Annotation guidance is captured in annotation_web_client/templates/annotation in the phaseTagging.html and responsibilityTagging.html files for cancer phases and patient responsibilities respectively.
Generally, each folder contains an independent analysis.
annotation_data - Utility functions for loading the CaringBridge data alongside the human annotations of that data.
annotation_web_client - Contains the code for the web tool used for all annotations.
classification - Contains all code related to classification, both ML and keyword based.
death_analysis - All code related to sampling additional EOL sites as described in the paper.
expert_validity - Analysis of survey administered to expert to check validity of codebook.
extract_site_features - Converts raw CaringBridge data to Sqlite or Feather format for easier downstream processing.
identify_candidate_sites - Code to identify the set of sites included in this study, to classify author type, and to generate annotation sets.
journey_phases - Random analyses related more to the phases, including IRR, etc.
patient_responsibilities - Random analyses related more to the responsibilities, including IRR, disagreement discussion, etc.