toolEnglish

Data Inputs and Source Review

How upload mode, URL preview, catalog metadata, descriptor files, AI interpretation, and CSV limits work.

EnglishEspaƱolFranƧais

Documentation follows the current system language. Missing translations fall back to the page's canonical language.

Data Inputs And Source Review

The Overview module is where a profile starts. It supports two source modes: Upload and URL.

CSV URL review sequence

For URL sources, the workbench previews the resource before profiling so users can verify the download target and metadata.

Upload Mode

Upload mode is best for local exports, internal extracts, or files that have already been reviewed outside Albarena. The selected file name is stored in the model data so later reviewers can see which file was profiled.

Before uploading, confirm:

  • the file is a CSV, not a spreadsheet workbook
  • the first row contains headers
  • sensitive fields are allowed in the workspace
  • the export represents the period and scope being discussed
  • the file size is within the group or organization CSV profiling limit

URL Mode

URL mode is best for open-data catalogs and hosted CSV resources. The tool requires a preview before profiling. The preview can resolve catalog pages, identify the CSV resource, show file size and content type, extract catalog metadata, show column descriptions when available, and list sample headers.

The user must confirm the preview before running the profile. This prevents accidental profiling of the wrong download target.

Metadata File Or Descriptor Text

A descriptor gives the profile business context. It can be a text, markdown, JSON, CSV, or plain notes file. The descriptor can also be entered as text.

Useful descriptor content includes:

  • dataset owner and source system
  • column definitions and units
  • collection period and refresh cadence
  • filters applied before export
  • known missing-value codes
  • category definitions
  • privacy or usage restrictions

AI Interpretation

When AI interpretation is enabled, the profile can produce narrative findings, recommendations, signals, next analyses, and story sections. When AI is unavailable or disabled, the tool still produces deterministic profiling metrics and fallback summaries.

Limits And Safety

CSV profiling limits can come from platform defaults, organization settings, or group overrides. URL preview shows the active limit source. Row caps may be applied for compatible CSV APIs so a very large public dataset can still be profiled safely.

Do not use URL mode to download data you are not authorized to access. Do not upload sensitive personal data unless the organization has approved that use.