Finding data — Collections and search

Before almost any analysis, the first question is "where does the data I need live?" Portal offers two patterns: browse and narrow down. Analysts switch between them constantly.

Pattern 1 — Browsing Collections

Click Collections in the left navigation to open the collection tree.

Resources have visually distinct icons:

• Dataset — Tabular data. View it row by row or wire it into a chart.
• Code — A Python or SQL snippet. Called from other stages.
• Pipeline — An execution flow that ties multiple steps together.
• Knowledge — Documents and search indexes used by AI and retrieval-augmented generation (RAG).
• Dashboard — A canvas of widgets.
• Agent — A resource that combines models and tools to answer questions or perform tasks.

Expanding folders is the right move when the question is "what does this team have?" Click a resource and the metadata card opens on the right, showing row count, last updated, and owner at a glance.

Pattern 2 — Narrowing with global search

Type into the search bar at the top (or press ⌘KCtrl+K⌘K / Ctrl+K). Every resource can be found by name or alias; depending on its type, its description or collection name may also be searchable. Relations additionally match tags and their connected endpoints, but tags on ordinary collections and datasets are not searchable. Results are grouped by resource type.

Two tips that make search more useful:

• Remembering part of the name is enough. The tokenizer handles both Korean and English.
• Use type filters to cut the noise. Limiting to "dashboards only" or "datasets only" keeps results focused when the same keyword shows up across resource types.

When search isn't returning what you expect, check in this order:

1. Search term — Try part of the resource's system name or visible alias. For types that search descriptions, such as dashboards and entities, you can also try a keyword from the description.
2. Type filter — If one resource type is selected, switch back to All and check again.
3. Permissions — If it still does not appear, confirm that you have permission to view the resource.
4. AI Assistant — If the steps above do not find it, ask in natural language (covered in the last lesson of this Path).

What to take away from this lesson

• Visually distinguish resource types in the collection tree
• Apply type filters in ⌘KCtrl+K⌘K / Ctrl+K search
• When results are empty, check the search term, type filter, and permissions in order

Next lesson

Now you'll create a collection of your own.