Published November 13, 2024
| Version v1
Dataset
Open
Jupyter Notebook Activity Dataset (rsds-20241113)
Description
List of data
- rsds-20241113.zip: Collection of SQLite database files
- image.tar.gz: Docker image provided in our data collection experiment
- redspot-341ffa5.zip: Redspot source code (redspot@341ffa5)
Extended version of Section 2D of our paper
Redspot is a Jupyter extension (i.e., Python package) that records activity signals. However, it also offers interfaces to read recorded signals. The following shows the most basic usage of its command-line interface:redspot replay <path-to-db>This command generates snapshots (.ipynb files) restored from the signal records. Note that this command does not produce a snapshot for every signal. Since the change represented by a single signal is typically minimal (e.g., one keystroke), generating a snapshot for each signal results in a meaninglessly large number of snapshots. However, we want to obtain signal-level snapshots for some analyses. In such cases, one can analyze them using the application programming interfaces:
from redspot import databasefrom redspot.notebook import Notebooknbk = Notebook()for signal in database.get("path-to-db"): time, panel, kind, args = signal nbk.apply(kind, args) # apply change print(nbk) # print notebookTo record activities, one needs to run the Redspot command in the recording mode as follows:
redspot recordThis command launches Jupyter Notebook with Redspot enabled. Activities made in the launched environment are stored in an SQLite file named ``redspot.db'' under the current path.
To launch the environment we provided to the participants, one first needs to download and import the image (image.tar.gz). One can then run the image with the following command:
docker run --rm -it -p8888:8888 <image-name>Note that the SQLite file is generated in the running container. The file can be downloaded into the host machine via the file viewer of Jupyter Notebook.