OneStop dataset

OneStop Eye Movements is a 360-participant English eye-tracking-while-reading corpus (Berzak, Malmaud, Shubi, Meiri, Lion, Levy, Scientific Data 2025, doi:10.1038/s41597-025-06272-2). The app's bundled demo is a 3-participant subset of it; this page covers loading the full public corpus from OSF as a public dataset.

Two ways to load OneStop

• Public dataset (this page) — the OneStop reports as a public dataset: the Public variant downloads from OSF on demand (no setup); the LaCC lab variant reads a local lab-processed export. Pick the variant, reading regime, and trial parts in the sidebar.
• OneStop server bundle — points at a local lacclab export via the $ONESTOP_DATA_DIR environment variable, with per-pid Parquet shards for review-app deep links. See Export & troubleshooting.

The corpus is 360 L1-English readers reading 30 Guardian articles (162 paragraphs, each in an Advanced and an Elementary version) — ~19.4k regular trials.

Loading it

OneStop is exposed as a Public dataset. In the app, choose Public datasets → OneStop, then in the ⚙️ Configure → Options group pick a Variant, a Reading regime, and one or more Parts:

Variant

Variant	What it is
Public (OSF download)	Reports fetched from OSF on demand, cached on disk.
LaCC lab (local export)	A lab-processed export with extra derived columns (unique_paragraph_id, span indices, normalized dwell, …). No download — point at your local folder (default is the lab OneDrive path, editable / ONESTOP_LACCLAB_DIR).

Reading regime

Regime	What it is
Ordinary reading	Standard paragraph reading.
Information seeking	Reading to answer a known question.
Repeated reading	Re-reading the same paragraphs.
Information seeking (repeated)	Information seeking during repeated reading.

Parts — which screen of a trial to load (default Paragraph):

Part (app label)	CLI / deep-link id	What it is
Title	Title	The article title screen.
Question preview	Question_Preview	The question shown before reading (information-seeking regimes).
Paragraph	Paragraph	The reading passage (the default).
Question	Questions	The question re-shown after reading.
Answers	Answers	The four answer choices.
Question + answers (QA)	QA	The combined question-and-answers screen.
Feedback	Feedback	The one-second correctness notification.

The first column is what the app's Parts picker shows; the second is the literal id for --onestop-part on the CLI and the ?onestop_parts= deep-link parameter.

Every part ships an interest-area report (one row per word, with bounding boxes and reading measures) and a fixation report, all in the same schema — so each part renders as a scanpath. Selecting several parts makes each part its own trial (the part is folded into the trial id, e.g. Paragraph::1 vs Title::1, so their word boxes don't collide). On OSF only Paragraph is regime-split; the other parts come from the all-regimes full release, so they load regardless of the chosen regime.

The ⚙️ Configure → Data location section lists the Expected files and shows whether they're already present. For the Public variant, if they're present the corpus loads with no network access; if not, click ⬇ Download to fetch them into the folder (cached on disk, so only the first load pays the download — reports range from tens to a few hundred MB each).

OneStop's reports use the same schema as the bundled demo, so they flow through the normal auto-detect → normalize pipeline — the Column mapping panels still appear and stay overridable. Fixation and interest-area coordinates are full-screen pixels on OneStop's 2560×1440 presentation monitor, so the canvas renders true-to-scale to that monitor.

From the Python API

The same loader is available headlessly — load_onestop returns normalized, plot-ready frames:

import scanpath_studio as sps

# Fetch (public variant) + normalize the chosen regime + parts (cached under root).
words, fixations = sps.load_onestop(
    "data/OneStop",
    regime="ordinary",
    parts=["Paragraph"],          # any subset of the seven parts
    variant="public",              # or "lacclab" for a local export
    download=True,                 # public variant only
)
fig = sps.plot_scanpath(words, fixations, canvas_size=(2560, 1440))

For the raw (pre-normalization) frames, use scanpath_studio.datasets.onestop_raw_frames(...) with the same arguments.

From the command line

scanpath-studio render --onestop data/OneStop \
    --onestop-regime ordinary --onestop-part Paragraph \
    -p <participant> -t <trial> -o out.html

--onestop-part is repeatable; --onestop-variant is public (default) or lacclab.

The implementation lives in datasets.py (onestop_raw_frames / load_onestop / download_onestop); the OSF file ids per regime + part come from the OneStop repo's download_data_files.py.