Saving and Loading¶
Your work can be saved as a local file or pushed to TriplyDB, and reopened later from either. The load and save buttons sit in the top bar of the header, so these actions are available on every tab. Each opens a menu with a Locally section (JSON, RDF) and a Remotely section (Triply).
Saving¶
You have three options:
| Save as | Result |
|---|---|
| JSON | Downloads the interpretation as a .json file with a timestamped name. This is the editor's native format and the fastest, most reliable round trip. |
| RDF | Converts the interpretation to RDF (via the wrap-up service) and downloads a .trig file — suitable for sharing as Linked Data. |
| Triply | Converts to RDF and uploads it to the TriplyDB knowledge graph. |
When saving to TriplyDB, only graphs that are not already present online are uploaded, so re-saving an interpretation will not create duplicates.
Save early, save often
Because the JSON format needs no conversion service, downloading a JSON file is the quickest way to checkpoint your work mid-interpretation.
Loading¶
You can reopen an interpretation in three ways:
- JSON — upload a previously saved
.jsoninterpretation. The editor then opens the Interpret sources tab. - RDF — upload a
.triginterpretation, which the unwrap service converts back into editor frames. The editor then opens the Interpret sources tab. - Triply — a dialog lists the tasks in TriplyDB (title, creator, and date). Select one and click Retrieve task; the editor pulls the task together with its sources. The current tab stays open, so switch to Interpret sources to continue the work.
Loading restores everything: the task details, the source documents (including which sentences were selected and which headings were collapsed), all frames and their roles, the boolean constructs, the text annotations and their underlining, and any comments.
Compatibility¶
The editor reads older interpretations as well as current ones. Interpretations that used a single fact subtype, that referenced the actor of a claim-duty under its old name, or that stored comments as plain strings, are all upgraded automatically on load — so you can safely reopen work created with earlier versions of the editor.