Skip to content
English - United Kingdom
Request Support
  • There are no suggestions because the search field is empty.

Background data mapping suggestions

Overview

XYCLE can identify missing background data in your LCA models and suggest ecoinvent datasets that match your inventory items. Generate suggestions across all unmapped inputs at once, then review candidates row by row.


Each mapping suggestion includes a confidence rating—strong, partial, or weak—and a one-line explanation of why it was chosen. You decide whether to approve, dismiss, or find your own mapping.

When to use mapping suggestions

Use mapping suggestions when:


  • You have inventory items with no background dataset mapped yet
  • You want to fill gaps faster than manual searching
  • You're modelling at scale and need to spot unmapped items quickly

Do not use suggestions when:

  • You already have a manually validated dataset selected

  • You need a foreground dataset (product-specific inventory)

  • Your process is so novel or specialised that no proxy will work well

  • You want automatic, hands-off data mapping—suggestions always require your approval

Suggestions

Suggestions come from ecoinvent and are based on four fields from your inventory item:


  • Item name—matched against ecoinvent product and activity names
  • Description (if present)—searched for key terms and product synonyms
  • Unit—used to filter candidates (e.g., kg, m, m³)
  • Region—if you mention a specific geography in the description, XYCLE prioritises regional data first, then falls back to global proxies

XYCLE returns up to three candidates per item. Each candidate shows the activity name, reference product, database, region, unit, confidence label and a one-line rationale.


N.b. Confidence reflects how closely the dataset aligns with your inventory item's name, unit, and region. A strong match means all three align well. A weak match means the dataset is a loosely related proxy.


Not currently covered:

  • Foreground data (your own processes, emissions, or waste streams)
  • Elementary flows (EF 3.1)
  • Automatic quality ratings (DQR, data quality indicators)
  • Supplier-specific or custom LCI datasets
  • Inputs other than material/energy inputs (M1)

How to use mapping suggestions

Generating suggestions

  1. Open a model and ensure that it includes at least one inventory item including item name.
  2. Click Generate suggestions in the INPUTS-DATASET column header. XYCLE runs across all unmapped input rows and populates candidates in the background.
  3. When generation is complete, unmapped rows with candidates show a View button and a suggestion count (e.g., "2 suggestions"). Rows where no candidate was found show "No suggestions found".

N.b. The Generate suggestions button is disabled if no database is selected in the background data filter, or if there are no unmapped input rows.

Reviewing candidates

  1. Click View on any row with suggestions.
  2. A panel slides in from the right showing up to three candidate cards.
  3. For each card, review:
    • The activity name and reference product
    • The database, region, and unit
    • The confidence label—strong (green), partial (amber), or weak (grey)
    • The one-line rationale explaining the match
  4. Inspect the dataset: Click Open (the icon on the right of each card) to open the Background Data panel for that dataset. There you can check the full process description, geographic assumptions, data vintage, and technology scope.
  5. Return to the candidate card and choose to approve or dismiss.

Approving a suggestion

  1. Click Approve on a candidate you trust.
  2. The mapping is applied immediately—no secondary confirmation is required.
  3. The row shows the mapped dataset name.

Dismissing a suggestion

  1. Click Dismiss on a candidate you don't want.
  2. The candidate is removed from view for that row.
  3. You can still search for and map an alternative dataset manually.
Limitations and gotchas

High confidence is not certainty

A "strong" confidence means the match is good on the fields XYCLE can evaluate—name, unit, and region. It does not mean the dataset is perfect for your context. Always spot-check:


  • The processes included in the dataset
  • The time period the data represents
  • Any geographic or technology assumptions in the process

No suggestions found

Some inventory items have no suitable candidate in ecoinvent:


  • Novel materials (e.g. emerging battery chemistries) without established processes
  • Highly specialised processes with no proxy
  • Regional variants not yet in ecoinvent

If XYCLE shows "No suggestions found", search ecoinvent manually, check external databases (e.g. Sphera, GaBi), or consider creating a custom dataset.

Weak confidence candidates

Weak means the candidate is the closest available but a real mismatch—different material class, very different scope, or a broad proxy. If accepting a weak match, document your reasoning in the item description or a model note.

Troubleshooting

"No suggestions found" appears

Reason: XYCLE found no candidates in ecoinvent for this inventory item.


Solution: Add more information to the inventory item or; search ecoinvent manually using the Search tool, or browse the Process Database directly. If no match exists, consider a proxy from a different region or technology, or create a custom dataset if time permits.

Suggestions look wrong or off-target

Reason: The inventory item name or description may be ambiguous or generic.


Solution:


  • Refine the item name or description to be more specific (e.g. "Steel, low-carbon, hot-rolled, EU" rather than "Steel").
  • Review the rationale on each card to spot mismatches.
  • Click Generate suggestions again after editing—suggestions are cached per item and regenerate when the name, description, or unit changes.
Best practices
  1. Review before approving: Even strong matches warrant a quick check of the full process record via the Open button.

  1. Iterate on item names: Suggestions are regenerated when item name, description, or unit changes. If the first round returns poor candidates, refine the item name to use ecoinvent vocabulary (e.g. "aluminium ingot" rather than "Al billet") and generate again.

  1. Audit before export: Before sharing or exporting your model, scan the inventory for rows that relied on suggested mappings and confirm each one matches your assumptions. An approved suggestion is still your responsibility.