Let's break down this Error 002—it's a parameter parsing failure, which means the Score Matchbox Recommender component isn't receiving input parameters or data that matches its expected format, even if you followed the example steps perfectly. Here are actionable troubleshooting steps to resolve this:

1. Verify Input Dataset Schema Consistency #

The Matchbox recommender is strict about input schema alignment between training and scoring phases:

• Double-check that your scoring dataset uses the exact same field names and data types as the dataset used to train the Matchbox model. For example, if your training data uses UserID (string type), your scoring input can't use User_Id (integer type) or any variation.
• Head to the "Schema" tab of both your training and scoring datasets to confirm field names, data types, and no missing required fields (like user IDs or item IDs).

2. Audit Score Component Parameter Configurations #

Dig into the Score Matchbox Recommender component's settings to catch misconfigured parameters:

• Ensure you've connected the correct trained Matchbox model to the component (not an unrelated model or an untrained one).
• Check parameter values for validity:
  • The number of recommendations to generate must be a positive integer (no zeros or negatives).
  • If you enabled "Exclude items already interacted by the user", make sure your input dataset includes historical interaction data for users.
• Confirm all input ports are properly connected (model input, user input, item input—whichever applies to your pipeline setup).

3. Clean Up Special Characters or Corrupted Data #

Odds are your dataset has hidden formatting issues that break parsing:

• Scan user IDs, item IDs, or text fields for special characters (commas, quotes, line breaks) that might confuse the component. Filter or escape these characters before feeding data into the pipeline.
• Remove blank rows or rows with malformed data—even a single corrupted row can trigger a parsing failure across the entire component.

4. Reset Pipeline or Reimport the Example Project #

Sometimes environment quirks or cached data cause unexpected issues:

• Try restarting your pipeline run and clearing any cached intermediate data (you can do this in the pipeline's "Run history" tab).
• If the problem persists, reimport the official "Recommender: Movie recommendation" example project from Azure ML's gallery—this ensures you're working with the default, validated component versions and configurations.

Quick Test to Isolate the Issue #

Swap your custom dataset with the sample dataset included in the example project. If the pipeline runs successfully, the problem is definitely in your dataset's format. If it still fails, focus on component version mismatches or workspace environment issues.

内容的提问来源于stack exchange，提问作者user9158446