Source: synthetic_gpt4o_demodbs (Neo4j text2cypher)
Total valid queries: 807
Products, orders, suppliers, customers, categories.
queries.csv: 807 converted queries- 0 failed queries
Total: 807 + 0 = 807 / 807 ✓
None.
Queries 66, 78, 98, 145, 148, 161, 167, 176, 261, 269, 308, 322, 375, 402, 470, 515, 520, 522, 634, 637, 672, 718, 767, 789 all required numeric operations on freight (comparison, sort, sum, avg). The original Neo4j dataset stored freight as a string and used toFloat() at query time. Fixed by changing the schema from attribute freight value string to attribute freight value double, which correctly models the semantic type.
Previously failed due to collect() + UNWIND for product co-occurrence. Converted using a self-join on the orders relation to find product pairs appearing in the same order, with $id1 < $id2 to deduplicate pairs.
Previously failed due to COLLECT() + REDUCE + RANGE + SIZE() for computing cumulative price variation. Converted using max_order_price() and min_order_price() custom functions on the order_unit_price attribute of the orders relation, computing variation as $max - $min.
Previously failed due to schema mismatch (no Employee entity or PROCESSED relation). Converted to return distinct employee_id values from orders placed by the specified customer, which is the closest available data in the schema.