JA EN
Manual / Chapter 3

Data Registration (Curation)

This chapter explains how to extract curves from figures in published papers using StarryDigitizer, attach composition and fabrication process information, and register the data. Any logged-in user can participate.

3.1 Registration workflow overview

Registration follows these 6 steps:

  1. Add the paper to a Database / My List (individually or in bulk by DOI)
  2. Click Data on the paper card to open the Data page
  3. Enable Edit mode and enter the Figure name and Caption
  4. Load the original plot image into StarryDigitizer and calibrate the axes
  5. Create one Dataset per sample, extract points → COPY TO CLIPBOARD
  6. Paste into XY-value, enter Sample information, property, and units, then click Save
Data page overview
Fig. 3-1: Data page layout. Top: Editors list and Figures / Samples tables. Bottom left: input form for Figure name / Caption / Sample information. Center: XY-value. Right: visualization plot.

3.2 Adding papers

Adding individually

  1. Select a My List and use + Add paper or the DOI input field.
  2. Entering a DOI automatically retrieves bibliographic information (title, authors, journal, volume, year, pages) from CrossRef.
  3. A unique SID (Starrydata ID) is assigned to that paper at the same time. The paper can be referenced by SID thereafter.
  4. If there are errors, manually correct them via Bibliographic information on the paper card.

Adding papers in bulk by DOI

  1. Paste DOIs into the Add papers from DOIs text box in the lower part of the left sidebar, one DOI per line (you can paste directly from an Excel column).
  2. Click Add Papers. Existing DOIs are simply added to the My List; new DOIs are looked up via CrossRef, registered, and assigned a new SID.
SID vs. sample_id: SID is a unique ID per paper (one assigned when a DOI is added). sample_id is a per-sample ID — many are generated within a single paper. The "SID" search field refers to the former.

3.3 Opening the Data page

Click the Data link on a paper card to open the Data page. The Data page is structured as follows (Fig. 3-1):

  • Header: Paper title, SID, DOI, number of Figures and Samples, links to Original paper / Bibliographic information / Data / Sample Table / Summary, etc.
  • Editors: List of Curator usernames who have edited this paper
  • Figures table: Figure ID / Figure name / Property(x), (y) / Unit(x), (y)
  • Samples table: Sample ID / Sample name / Composition
  • Edit form (bottom left): Figure name / Caption / Sample name / Composition / Property(x), (y) / Unit(x), (y) / Measurement conditions / Comments / Save button
  • XY-value (center): Area where you paste extracted numerical data
  • Visualization plot (right): Plotly display of registered or currently entered data

3.4 Edit mode and creating Figures / Samples

Toggle Edit in the upper right to ON — the form background turns blue and editing is enabled.

  1. Figure name: Enter the same notation as in the paper (e.g., Fig 5a, 4(a), S3) and press Enter to confirm. This field cannot be changed afterwards.
  2. Caption: Copy and paste the figure caption from the paper.
    • Prohibited characters: < > " ' &.
    • Special characters such as Ω and µ cannot be registered as-is — replace them with alphabetic equivalents (e.g., ommega, mu).
  3. Sample name: Enter the name matching the plot legend (e.g., y=10%, 1h BM-SPS) and press Enter to confirm.
  4. Composition: Chemical composition formula (e.g., Mg3Bi0.33, Bi2Te2.85Se0.15).
  5. Property(x), Property(y): Select the property from the dropdown. Toggle between Recommended (domain standard) and All (all options) using the radio buttons.
  6. Unit(x), Unit(y): Specify units in SI base + prefix/exponent notation (e.g., 10^0 × °C, 10^0 × VK^-1).
Note: For safety, follow the templates of the domain-specific Project (e.g., ThermoelectricMaterials) for Property / Unit. In GeneralDB the candidate list is very broad, so registration in a domain-specific DB is recommended for better analytical usability.

3.5 Extracting curves with StarryDigitizer

Launch StarryDigitizer from the right panel of the Data page (or a dedicated tab on the paper page). There are two tabs — WebPlotDigitizer and StarryDigitizer — and this manual covers the newer StarryDigitizer.

Example of the source plot to extract (S vs T, y=10%–42%)
Fig. 3-2: Example of the source plot to extract (paper Fig 5a, S vs T, 6 series: y=10%/20%/28%/38%/40%/42%). Take a screenshot in advance from the PDF or HTML (Mac: cmd+shift+4 / Win: win+shift+s).
Paper used in this demo:
F. Garmroudi, M. Parzer, A. Riss, N. Reumann, B. Hinterleitner, K. Tobita, Y. Katsura, K. Kimura, T. Mori, E. Bauer, Solubility limit and annealing effects on the microstructure & thermoelectric properties of Fe2V1−xTaxAl1−ySiy Heusler compounds, Acta Materialia 212 (2021) 116867. DOI: 10.1016/j.actamat.2021.116867
StarryDigitizer immediately after launch
Fig. 3-3: StarryDigitizer launch screen. Load an image by pasting or selecting a file via Choose an image in the upper left. XY Axes 1 is created automatically in the XY Axes List.

3.5.1 Axis calibration (Set Axes)

  1. Enter the known coordinate values for two points each on the x and y axes in the x1 / x2 / y1 / y2 input fields in the left sidebar (e.g., x1=30, x2=90, y1=-60, y2=120).
  2. For logarithmic axes, enable the Log checkbox.
  3. Choose 2 Points / 4 Points in Calibration mode and keep Show axis marker ON to confirm positions visually.
  4. Click the corresponding positions on the plot, then click SET AXES to complete calibration. Click CLEAR XY AXES to start over.

3.5.2 Creating Datasets

  1. In the Datasets section of the left sidebar, click + to add as many Datasets as there are samples (e.g., 6 samples → datasets 1–6).
  2. With each Dataset selected, choose the matching series color from the Color palette on the right (match the legend colors).

3.5.3 Extracting data points

The basic approach is to use the Automatic Extraction section in the right panel.

  1. Algorithm: Choose Line Extract (for line plots) or Averaging Window (for scatter plots).
  2. ΔX / ΔY: Specify the extraction interval (px). Use smaller values for dense points, larger for sparse.
  3. Restrict the extraction area with PEN / BOX in Selection Area, then click RUN to auto-detect points.
  4. Remove unwanted points with ERASER, clear all with CLEAR. Use ADD (A) / EDIT (E) / DELETE (D) in Manual Extraction to add or correct individual points.
  5. Extracted points are immediately reflected in the X / Y table at the bottom left.
StarryDigitizer with calibrated axes and extracted points
Fig. 3-4: Extracted point cloud on the calibrated plot. Switch the Color palette for each series and repeat RUN → extract → verify for each of the 6 Datasets.

3.5.4 Copying to clipboard

Click the COPY TO CLIPBOARD button at the bottom left to copy the (X, Y) pairs of the selected Dataset to the clipboard as tab-delimited text.

3.6 Pasting into XY-value and saving

  1. Return to the Data page and enable Edit mode for the target Sample row.
  2. Click in the XY-value area in the center and paste with Ctrl/Cmd + V. The right-side plot updates immediately, allowing you to compare with the original figure.
  3. Confirm that all required fields are filled (Figure name / Sample name / Composition / Property / Unit).
  4. Click Save to complete registration. New rows will appear in the Figures / Samples tables.
Edit form, XY-value area, visualization plot, and Save button
Fig. 3-5: The edit mode form. Enter the figure caption in Caption, composition in Composition, properties in Property(x)/(y), units in Unit(x)/(y), and Measurement conditions (Temperature / Pressure / Other / Magnetic Field / Orientation) as needed. Confirm the right-side plot matches the original figure before clicking Save.
Automatic unit conversion: All values are automatically converted to SI units (K, V/K, etc.) upon saving. The display scale may appear temporarily off right after saving, but reloading the page will show the correct scale.

3.7 Editing existing data

Anyone can edit and correct data registered by other users. This is the core of Starrydata2's collaborative data curation.

  1. Click a Sample row to load it into the edit form.
  2. Enable the Edit toggle in the upper right and make changes (redo XY-value, correct composition, delete erroneous points, etc.).
  3. Click Save to confirm. Your Username will be appended to the Editors table.

3.8 Best practices for registration

  • Figure name must strictly follow the paper's notation (e.g., Fig 5a, 4(a), S3).
  • Sample name should match the plot legend exactly (e.g., y=10%, 1h BM-SPS). This prevents the accidental loss of sample identification if the composition alone is ambiguous.
  • Even within the same figure, curves with different x-axis properties (e.g., temperature dependence vs. carrier concentration dependence) should be registered as separate Figures.
  • Be careful of prohibited characters < > " ' & and special characters such as Ω µ (→ use ommega / mu).
  • If you plan to curate actively for more than a month, contact the Starrydata team to be introduced to the question channel (inquire via Contact).

Next steps

Reference