Using a lab workbench

Most labs open in a shared workspace. The views, filters and wording depend on the lab and its data; this page describes the controls they have in common.

Workspace layout

• View strip — tabs across the top. Summary is always first; the other tabs are the views the lab offers, for example People of interest, Graph, Timeline or Statements.
• Main area — the active view.
• Sidebar on the right — search, filters, export and the working set. On narrow screens it opens from the Lab controls button in the bottom-right corner.
• Advisor switch at the right end of the view strip — opens and closes Lab Advisor.

The breadcrumb bar above the workspace shows the active view and, where applicable, the selected group, the focused entity or the open dossier. Selecting an earlier item in the bar returns to it.

Summary

Summary describes the release before its contents: the key counts under At a glance, the period covered, the relationship types, the most cited sources and publishers, what was deliberately left out, and the details of the release. In labs built on statements, the summary describes the statements instead — their number, themes and activity over time.

Search and filters

• Search this release finds entities by name. Type at least two characters and select a result to focus it.
• When the release is divided into groups, a group selector narrows the views to one group; All groups returns to the whole release.
• Research status switches show or hide relationships with a given status.

In labs built on statements, the sidebar offers Search statements and filters such as theme, statement form and Date range. Clear filters resets them.

Graph

Graph shows entities and their relationships. Relationship lines are coloured by research status; the legend sits at the bottom left.

• Select an entity to focus it. The panel at the top left shows its name (or Network overview when nothing is focused) and how many entities and relationships are visible.
• Keep adds the focused entity to the marked set and Remove from set takes it out; Set aside hides it from the view. ⌘/Ctrl-click marks several entities without moving the focus.
• +, − and Fit zoom the graph. When only part of the release is visible, Regroup lays out the visible entities on their own, and Full map returns them to their places on the full map.
• Select a relationship line to open its evidence.

Some relationships appear only after zooming in; the legend says so. These controls change only the view — they never add, remove or rewrite relationships in the release.

Working set

The Working set section of the sidebar appears when entities are marked or set aside, or when the view is limited to some entity types:

• marked entities are listed by name — select a name to focus it or × to unmark it; Clear unmarks all;
• entities that were set aside are counted; Restore all brings them back;
• an active filter by entity type is shown with Show all, which removes it.

People of interest

People of interest lists the people in the release with an Interest rating, their group, their relationships and the latest date. Selecting a row opens the person's dossier. Selecting People of interest in the breadcrumb bar returns to the list.

Timeline

Timeline builds the chronology of the focused entity. Open an entity from the graph or the people list first. Each dated entry can show its source or target in the graph. In labs built on statements, the timeline places statements on a time axis, grouped by the option chosen in Group timeline by.

Dossiers and evidence

A dossier opens with a Summary of the person and then groups their reviewed relationships by type. Each relationship lists its sources. When the release stores an archived copy of a source, Open document displays it in place.

Selecting a relationship line in the graph opens Relationship evidence: the status, period, publisher and source of the relationship, Open source, and any Further evidence. When the release does not store a snapshot of a source, the source can be cited but not opened. A relationship whose evidence is unavailable should not be inferred from the graph alone.

Findings from the open web

A dossier can include an Open web section with findings collected from the open web. They are not part of the release and have not been reviewed. Each finding shows its source site, an excerpt, what it was matched on, and how many independent texts and sites support it.

• Accept marks the finding for the next release; it then reads Accepted for the next release.
• Reject marks the finding as Rejected.
• Undo returns a decided finding to undecided.

Accepting a finding never adds it to the release that is currently open.

Export

The Export section of the sidebar downloads an Excel workbook (.xlsx) for the current scope. The label shows what will be exported:

• Export whole release — nothing is selected;
• Export followed by the group name — a group is selected;
• Export with the number of marked entities — entities are marked.

The workbook includes only the research statuses that are currently displayed. Labs built on statements offer Export statements and Export timeline instead. If a download fails, the control reports the failure and can be selected again. Export requires an active session with access to the lab.

Lab Advisor

Lab Advisor is the Advisor working inside the lab. The switch at the right end of the view strip opens it next to the view. On wide screens the panel can be resized by dragging its edge; on narrow screens it covers the view. Persate remembers whether the panel was open.

• Each question is sent together with what is on screen — the active view, the focused entity or open dossier, marked entities and active filters — so questions can refer to this group or the marked people.
• Lab Advisor answers from the lab's release. It can also mark entities, move the view and prepare a download, which appears in its answer.
• Stop cancels the running answer. New conversation clears the panel and starts a new conversation.
• If the connection drops during an answer, Reconnect resumes it. If an answer stops early, the panel says why; a narrower follow-up question usually completes it.

Empty and error states