config/codebert_labels.yaml

# Primary labels for CodeBERT cross-encoder training
labels:
  - JOIN_ERROR
  - AGGREGATION_ERROR
  - FILTER_ERROR
  - WINDOW_FUNCTION_ERROR
  - SUBQUERY_ERROR
  - NULL_HANDLING_ERROR
  - PERFORMANCE_ERROR
  - LOGICAL_ERROR
  - SYNTAX_ERROR

# Map dataset label_name values → one or more CodeBERT labels (multi-label)
alias_map:
  JOIN_ERROR: [JOIN_ERROR]
  AGGREGATION_ERROR: [AGGREGATION_ERROR]
  HAVING_WHERE_ERROR: [AGGREGATION_ERROR]
  FILTERING_ERROR: [FILTER_ERROR]
  WINDOW_FUNCTION_ERROR: [WINDOW_FUNCTION_ERROR]
  SUBQUERY_ERROR: [SUBQUERY_ERROR]
  NULL_HANDLING_ERROR: [NULL_HANDLING_ERROR]
  PERFORMANCE_ERROR: [PERFORMANCE_ERROR]
  LOGICAL_QUERY_ERROR: [LOGICAL_ERROR]
  SYNTAX_ERROR: [SYNTAX_ERROR]
  DATE_FUNCTION_ERROR: [SYNTAX_ERROR]
  COLUMN_REFERENCE_ERROR: [SYNTAX_ERROR]
  TABLE_REFERENCE_ERROR: [SYNTAX_ERROR]
  DATA_TYPE_ERROR: [SYNTAX_ERROR]
  DUPLICATE_RECORD_ERROR: [FILTER_ERROR]