Event-Based Analysis

Overview

Event-Based Analysis finds which entities were present in a geographic area during specific time windows. You draw one or more event polygons on a map, set a time range for each event, select an analysis preset, and run the analysis to discover subjects such as people, devices, or vehicles. Outputs include a paginated subject list with reasoning chains, plus optional subject position history on a map and timeline. Results depend on configured reasoning paths and require review before you treat presence as confirmation.

---

When to Use This Application

• You need to identify who or what was present at a location during a defined time window.
• You need to compare multiple events and find subjects present in all of them.
• You need to run a preset investigation strategy (for example “Find People” or “Find IMEIs”) without writing queries.
• You need to save events and results to a Workspace so other analysts can review them.
• You need map and timeline views for detections tied to a reasoning chain.

---

Before You Begin

• Ensure position-style geo-event data is available in Octostar for the area and time windows you plan to analyze.
• Confirm you have access to the destination Workspace folder where you will save event files and results.
• Confirm presets are configured and available. Presets are defined in config.yaml.
• If you plan to visualize subject positions after analysis, avoid changing config.yaml paths between running and visualizing the analysis.

---

Step-by-Step Walkthrough

Step 1 — Choose create mode or load mode

Use the Start Analysis section in one of two modes:

• Create: define new events by drawing polygons and setting time ranges.
• Load: load saved events from .ebas files stored in a workspace folder.

---

Step 2 — Create events (create mode)

Step	Action	Details
1	Draw a polygon on the map	Use the drawing tools to draw one or more polygons for the event.
2	Set the time range	Select start date/time and end date/time.
3	Select a preset from the dropdown	Presets define which tables and reasoning paths the analysis uses.
4	Preview the event	Select the preview action to see how many positions match your polygon and time range.
5	Add the event	Enter an event name and add it to the event list.

Repeat these steps to create multiple events.

---

Step 3 — Load events (load mode)

Load events when you already have .ebas files saved in a workspace.

Step	Action	Description
1	Browse workspace folders	The app lists folders that contain .ebas files.
2	Select a folder and load its events	Loaded events appear in the same event list used in create mode.

After loading, run the analysis using the same steps as create mode.

---

Step 4 — Run the analysis

Select one or more events from the event list and select Run. The app can run analysis in two ways:

• In-app execution with real-time progress streaming
• Background execution as a job Use in-app execution when you want to monitor progress live. Use background execution for longer runs.

---

Step 5 — Visualize results

Open the Visualize Results section to explore the output.

Load an analysis to visualize

Use the dropdown at the top of the page to choose what to visualize:

• In-app results from the most recent run
• Saved .eba files found in your workspaces

Browse subjects

After the analysis loads, the app shows a paginated list of subject cards. Each card includes:

• Label and type (subject label and concept type)
• Event chips (events where the subject appears)
• Tags (classification tags)
• Identifiers (raw identifiers linked to the subject)
• Reasoning chain (ontology path from geo-event to subject) Use pagination controls to move through results.

Filter subjects

Use filters to narrow the subject list:

• Event selection using event chips
• Subject concept type
• Identifier concept type
• Label regex
• Identifier label regex
• Event label regex
• Reasoning regex Event chip filtering shows only subjects present in all selected events.

View subject positions (map and timeline)

Expand a subject card to load position history:

• Map view showing the geo-event locations where the identifier was detected
• Timeline chart showing detections over time Position retrieval depends on paths in config.yaml. If the analysis reasoning chains do not match current paths, the app shows a warning and does not load positions.

---

Understanding the Output

Event-Based Analysis produces:

• Event list
• A list of named events, each defined by one or more polygons, a time range, and a preset.
• Subject list
• A paginated list of discovered subjects (people, devices, vehicles, and other concept types). Each subject includes identifiers and a reasoning chain.
• Event chips
• Colored indicators that show which events a subject appears in. Use chips to filter to subjects shared across events.
• Reasoning chain
• The ontology path used to connect a raw geo-event to the subject (for example position_of_identifier → phone_number → is_owned_by → legal_person). This explains why the subject appears in results.
• Subject position history
• A map and timeline for detections, when path matching succeeds. If config.yaml paths change after analysis, position history may not load.

---

Saving and Exporting Results

You can persist both events and analysis results to a workspace folder.

• Save events as .ebas files
  • What it does: saves each event as an individual .ebas file
  • Where it ends up: selected Workspace folder
  • Settings: destination folder path
• Save analysis results as a .eba archive
  • What it does: saves analysis output as a single .eba file for later visualization
  • Where it ends up: selected Workspace folder
  • Settings: destination folder path, filename

---

Tips for Best Results

• Create events with precise polygons. Large polygons increase noise and reduce interpretability.
• Use separate events for separate places or time windows, then use event chips to find intersections.
• Preview each event before you add it to confirm the time window and polygon match expected record counts.
• Select presets that match your investigative goal. Presets control which tables and reasoning paths are active.
• Keep config.yaml paths stable between running an analysis and visualizing positions from that analysis.
• Use regex filters to narrow subject lists when labels vary in formatting.

---

Known Limitations

• Results depend on presets and reasoning paths defined in config.yaml. Misconfigured paths reduce recall or produce incomplete subject sets.
• Subject position history can fail to load when config.yaml paths change after analysis, because stored reasoning chains no longer match.
• Presence in results indicates a match through configured reasoning paths. It does not confirm identity without supporting evidence.
• Large event lists and broad presets can produce very large subject sets, which may reduce usability without filtering.