By Aini Putkonen, Aurélien Nioche, Markku Laine, Crista Kuuramo & Antti Oulasvirta
Copyright (c) Aalto University
This repository accompanies the paper "Fragmented Visual Attention in Web Browsing: Weibull Analysis of Item Visit Times", published at ECIR '23.
Create virtual environment:
python3 -m venv .env
Activate virtual environment:
source .env/bin/activate
Install required packages
python3 -m pip install -r requirements.txt
Deactivate virtual environment
deactivate
The folder paths are configured in CONFIG.py. The data is read from data/01-fit-data/filter=True by default (make sure to add the data file here). The columns in the data can be interpreted as follows:
| Name | Definition |
|---|---|
| y | visit time |
| N | # observations |
| i | indicator for picture |
| d | indicator for description |
| K | number of groups |
| x | group indicator |
Note that data is given for three foveal areas (number at the end of the file name indicating radii). To change which file is read, change RADIUS in CONFIG.py.
Model fitting can be run using the following command:
python3 -m src.fit
This will also take a couple of minutes. Results of model fitting are saved as a text file to data/02-results/fitting_results_separate.csv and data/02-results/fitting_results_separatecovariate.csv.
The results can be examined using the notebook notebooks/stan-plots. The produced plots are saved to data/02-results. Figures 3 and 4 from the paper are re-produced here.
Note that some of the variables in the repository are named differently to the paper. For instance, the shape k=alpha and the scale sigma=lambda to convey to the Stan documentation. The covariate for picture p=i, as to not confuse it with an index in the written paper. The Stan code for the model is saved in src/models/weibull.py.
TBD
If you have any questions, do not hesitate to contact the corresponding author at aini.putkonen@aalto.fi.