{ "captured": "2026-09-28", "environment": { "git_commit": "70debd55f5041b191bf79a3213cfcba5cae5abc0", "cuvs_commit": "c1414f7b36f21d98968900be2a46411e63e607b4", "server_binary": "target/release/algeon_server", "server_sha256": "bf9a3d1142d99cb238e0b034ac2c99cc23e7fb778dd05722e6f03942b28afc10", "cuvs_library": "target/native/cuvs-tests-build/libcuvs.so", "cuvs_sha256": "790bd2de19f420baca30ae1a8510f1a388943dd392cbb04b7f5dc5c33163af5f", "ld_library_path": "target/native/cuvs-tests-build:target/native/cuvs-tests-build/c", "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition, 595.91.07, 97887 MiB", "source_config": "docs/research/amazon_reviews_2023/server.toml", "notes": "Release rebuilt with the cuDF empty nested gather fix on the recorded base commit. Source hashes are in cudf_fix_sources.json. The verified cuVS test library matches the pinned fork.", "local_fix_sources": { "components/cudf/algeon-cudf/src/frame/dataframe/row_ops.rs": "5e08f7b13e86fd6a1d2f5919afccbd2d842d7890b60036036c595824487f79e7", "components/cudf/algeon-cudf/src/frame/series/context.rs": "daa656111b706a8123fa885914f3d2abc76dc9c2bee7c363f913b515aaa58114", "components/cudf/algeon-cudf/tests/frame/dataframe/gather_tests.rs": "a7f893de888e1111e3c134b89d52f78677785e69164d6c8218f0cc8efd4f5746" } }, "source": { "dataset": { "category": "Electronics", "source": "McAuley-Lab/Amazon-Reviews-2023 (https://huggingface.co/datasets/McAuley-Lab/Amazon-Reviews-2023)" }, "counts": { "max_timestamp_ms": 1694625981867, "min_timestamp_ms": 848336280000, "products": 1610012, "products_with_price": 526449, "products_without_reviews": 152, "reviews": 43886944, "users": 18286191, "verified_reviews": 40546884 }, "raw_checksums": { "metadata": { "bytes": 5246144134, "sha256": "5eb4dabd539586c49654297319ef178b62a2fe8e43108376feda125f04455d93" }, "reviews": { "bytes": 22616233652, "sha256": "bf11cf9c94c23ead7d3ffb174d18231fad8fcd76e72ad167f538a5599a38de8e" } } }, "embedding": { "model": "BAAI/bge-m3", "revision": "5617a9f61b028005a4858fdac845db406aefb181", "dimensions": 1024, "max_sequence_length": 8192, "inference_dtype": "float16", "stored_dtype": "float32", "normalization": "L2 in float32 after encoding", "batch_size": 32, "model_load_s": 2.957023192953784, "text_preparation": "full fields, HTML removed, entities decoded, whitespace collapsed", "packages": { "torch": "2.14.0", "sentence-transformers": "6.1.0", "transformers": "5.17.0", "numpy": "2.5.3", "pyarrow": "25.0.1", "duckdb": "1.5.5", "faiss-cpu": "1.15.1" }, "device": "NVIDIA RTX PRO 6000 Blackwell Server Edition", "runs": [ { "name": "intents", "rows": 14, "elapsed_s": 0.38415092800278217, "matrix_sha256": "82b537915f3dc07ff493e12492982e6ec5240a11ddd7f5d448af9e2582a9d59d", "matrix_bytes": 57344, "parquet_bytes": 58506, "torch_peak_allocated_bytes": 1176779776 }, { "name": "products", "rows": 1010000, "elapsed_s": 1797.4371816050261, "matrix_sha256": "bbaf20f08c19274a8e2fee91a44c3a4b9e2f18ea62b6f5c07ee204d9525a9c62", "matrix_bytes": 4136960000, "parquet_bytes": 4146868501, "torch_peak_allocated_bytes": 13788288000 }, { "name": "reviews", "rows": 1000000, "elapsed_s": 455.4924460509792, "matrix_sha256": "1e7190d5be2064fd6ec1cd1cec4baa22932901d821048df4d9f6107342678b66", "matrix_bytes": 4096000000, "parquet_bytes": 4106518425, "torch_peak_allocated_bytes": 7426816000 } ] }, "input_verification": { "products": { "rows": 1010000, "dimensions": 1024, "unique_ids": true, "parquet_matches_npy_and_source_order": true, "finite_and_non_null": true, "max_unit_norm_error": 1.7881393432617188e-07, "matrix_sha256": "bbaf20f08c19274a8e2fee91a44c3a4b9e2f18ea62b6f5c07ee204d9525a9c62", "row_groups": 8 }, "reviews": { "rows": 1000000, "dimensions": 1024, "unique_ids": true, "parquet_matches_npy_and_source_order": true, "finite_and_non_null": true, "max_unit_norm_error": 1.1920928955078125e-07, "matrix_sha256": "1e7190d5be2064fd6ec1cd1cec4baa22932901d821048df4d9f6107342678b66", "row_groups": 8 }, "intents": { "rows": 14, "dimensions": 1024, "unique_ids": true, "parquet_matches_npy_and_source_order": true, "finite_and_non_null": true, "max_unit_norm_error": 5.960464477539063e-08, "matrix_sha256": "82b537915f3dc07ff493e12492982e6ec5240a11ddd7f5d448af9e2582a9d59d", "row_groups": 1 } }, "timing_boundary": "Three sequential statements in one release server, existing vectors, reusable OS file cache. Includes planning, index build or fit, and complete Flight transfer; excludes embedding preparation and writing result files.", "verification": { "travel_shortlist": { "eligible_products": 117, "rows": 10, "duckdb_history_and_float64_knn_match": true }, "headphone_emerging": { "rows": 10, "duckdb_denominators_match": true, "examples_match_original_rows": true }, "headphone_umap2": { "rows": 9967, "cohort_ids_match": true, "finite_coordinates": true, "trustworthiness_1500_at_15": 0.7961367035281728, "repeat_max_coordinate_deltas": [ 0.0, 0.0 ] } }, "exact_batch_verification": { "recall_at_10_ids": 0.99972, "fraction_within_exact_top10_radius_plus_1e_5": 1.0, "max_reported_distance_error": 0.00022016573353201484, "mean_reported_distance_error": 6.306285734521012e-07, "reported_zero_with_true_distance_gt_1e_5": 3, "true_distance_rank_inversions_gt_1e_5": 0, "corpus_membership_ok": true, "queries_disjoint_from_corpus": true }, "queries": { "travel_shortlist": { "runs": 3, "statuses": [ "ok", "ok", "ok" ], "median_s": 1.544123484985903, "min_s": 1.441004102001898, "max_s": 2.4794984739855863, "rows": 10, "observations": [ { "query_id": "c3c662bc-f3cf-4089-95a1-c7809baef565", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "1885571400", "gpu_resident_input_handoffs": "3", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "1", "device_to_host_materialized_bytes": "2438", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "f43c0830-a794-42c8-bb97-3912dfc795dc", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "1885571400", "gpu_resident_input_handoffs": "3", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "1", "device_to_host_materialized_bytes": "2438", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "304c7133-67c9-4407-a689-8104dd5d0810", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "1885571400", "gpu_resident_input_handoffs": "3", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "1", "device_to_host_materialized_bytes": "2438", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] } ], "sql": "WITH review_history AS (\n SELECT product_id, COUNT(*) AS reviews_2022,\n AVG(CASE WHEN rating <= 2 THEN 1.0 ELSE 0.0 END) AS low_rating_share\n FROM reviews\n WHERE review_time >= TIMESTAMP '2022-01-01'\n AND review_time < TIMESTAMP '2023-01-01' AND verified_purchase\n GROUP BY product_id\n), eligible AS (\n SELECT v.id, v.vector\n FROM semantic_products_vectors v\n JOIN products p ON p.product_id = v.id\n JOIN review_history h ON h.product_id = v.id\n WHERE v.sample_rank < 1000000\n AND p.category_l3 = 'Headphones & Earbuds'\n AND p.price BETWEEN 20 AND 150\n AND h.reviews_2022 >= 100 AND h.low_rating_share <= 0.20\n)\nSELECT n.rank, p.product_id, p.title, p.price,\n h.reviews_2022, h.low_rating_share,\n 1.0 - n.distance / 2.0 AS cosine_similarity\nFROM cuvs_brute_force_knn(\n dataset => (SELECT id, vector FROM eligible),\n queries => (SELECT id, vector FROM intent_queries WHERE id = 0),\n k => 10, metric => 'l2_expanded') n\nJOIN products p ON p.product_id = n.neighbor_id\nJOIN review_history h ON h.product_id = p.product_id\nORDER BY n.rank\n", "sql_sha256": "1a910d84cf7f4e888da69618a84fe6eeb8af990f089c5f11605fc85f802f0c38", "illustrated_run": "c3c662bc-f3cf-4089-95a1-c7809baef565", "illustrated_rows": [ { "rank": 1, "product_id": 1444831, "title": "Srhythm NC35 Noise Cancelling Headphones Wireless Bluetooth 5.0, Fast Charge Over-Ear Lightweight Headset with Microphones,Mega Bass 50+ Hours’ Playtime", "price": 69.99, "reviews_2022": 353, "low_rating_share": 0.16147308781869688, "cosine_similarity": 0.6391699910163879 }, { "rank": 2, "product_id": 1484401, "title": "VELKPRO Wireless Headset with Microphone - Noise Canceling Headphones with Mic - On-Ear Earphones with USB Dongle for Office Work, Video Calling, Call Center - 32H Talk Time, Clear Audio Transmission", "price": 29.99, "reviews_2022": 103, "low_rating_share": 0.05825242718446602, "cosine_similarity": 0.6105114817619324 }, { "rank": 3, "product_id": 1562005, "title": "HROEENOI Active Noise Cancelling Wireless Bluetooth Over-Ear Headphones, Memory Foam Ear Cups, Quick Charge for 40H Playtime, Ideal for Travel, Home Office, Gym Workouts-Black", "price": 45.99, "reviews_2022": 354, "low_rating_share": 0.1751412429378531, "cosine_similarity": 0.6033956408500671 }, { "rank": 4, "product_id": 1557567, "title": "Soundcore by Anker Space Q45 Adaptive Active Noise Cancelling Headphones, Reduce Noise by Up to 98%, 50H Playtime, App Control, LDAC Hi-Res Wireless Audio, Comfortable Fit, Clear Calls, Bluetooth 5.3", "price": 149.99, "reviews_2022": 179, "low_rating_share": 0.11731843575418995, "cosine_similarity": 0.5999435186386108 }, { "rank": 5, "product_id": 1465416, "title": "PurelySound E7 Active Noise Cancelling, Wireless Over Ear Bluetooth Headphones, 20H Playtime, Rich Deep Bass, Comfortable Memory Foam Ear Cups for Travel, Home Office -Matte Black", "price": 39.48, "reviews_2022": 310, "low_rating_share": 0.14193548387096774, "cosine_similarity": 0.5994947552680969 }, { "rank": 6, "product_id": 1605871, "title": "Wireless Earbuds, Bluetooth Earbuds Environmental Noise Cancellation 4 Mic Call Noise Cancelling Mini Earbuds, Bluetooth 5.2 Light Weight Deep Bass Headphones for Work, Home Office", "price": 35.99, "reviews_2022": 409, "low_rating_share": 0.08312958435207823, "cosine_similarity": 0.5962716341018677 }, { "rank": 7, "product_id": 1412635, "title": "EarFun Air Pro 2 Wireless Earbuds, [2023 Version] Hybrid Active Noise Cancelling Wireless Earphones, Bluetooth 5.2 Headphones with 6 Mics, in-Ear Detection, App for Custom EQ, Wireless Charging, 34Hrs", "price": 69.99, "reviews_2022": 478, "low_rating_share": 0.1087866108786611, "cosine_similarity": 0.5941441059112549 }, { "rank": 8, "product_id": 1603005, "title": "Wireless Earbuds, Bluetooth 5.3 Headphones in Ear with Noise Cancelling Mic, Bluetooth Earbuds Stereo Bass, IP7 Waterproof Sports Earphones, 32H Playtime USB C Charging Ear Buds Purple for Android iOS", "price": 21.99, "reviews_2022": 960, "low_rating_share": 0.09166666666666666, "cosine_similarity": 0.5803655982017517 }, { "rank": 9, "product_id": 1125873, "title": "Soundcore Anker Life Q20 Hybrid Active Noise Cancelling Headphones, Wireless Over Ear Bluetooth Headphones with 40H Playtime, Hi-Res Audio, Deep Bass, Memory Foam Ear Cups and Headband (Renewed)", "price": 44.99, "reviews_2022": 117, "low_rating_share": 0.1623931623931624, "cosine_similarity": 0.5801839828491211 }, { "rank": 10, "product_id": 1339800, "title": "Soundcore by Anker Life Q30 Hybrid Active Noise Cancelling Headphones with Multiple Modes, Hi-Res Sound, Custom EQ via App, 40H Playtime, Comfortable Fit, Bluetooth Headphones, Multipoint Connection", "price": 79.99, "reviews_2022": 1508, "low_rating_share": 0.14655172413793102, "cosine_similarity": 0.579837441444397 } ] }, "catalog_batch": { "runs": 3, "statuses": [ "ok", "ok", "ok" ], "median_s": 2.5841987360036, "min_s": 2.5529906740412116, "max_s": 2.5986024220474064, "rows": 100000, "observations": [ { "query_id": "8850d8ec-b120-45b7-adf3-88377fccbaed", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "8306240624", "gpu_resident_input_handoffs": "4", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "16", "device_to_host_materialized_bytes": "31449135", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "600a3a6d-f1bf-4f4a-be4b-4c9eab0964ad", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "8306240624", "gpu_resident_input_handoffs": "4", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "16", "device_to_host_materialized_bytes": "31449135", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "bcdcacb4-aaca-4771-ba6a-0cedb96ff914", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "8306240624", "gpu_resident_input_handoffs": "4", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "16", "device_to_host_materialized_bytes": "31449135", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] } ], "sql": "SELECT n.query_id, q.title AS source_title, n.rank,\n n.neighbor_id, p.title AS candidate_title,\n 1.0 - n.distance / 2.0 AS cosine_similarity\nFROM cuvs_brute_force_knn(\n dataset => (SELECT * FROM product_corpus_1000000),\n queries => (SELECT * FROM product_queries_10000),\n k => 10, metric => 'l2_expanded') n\nJOIN semantic_products_text q ON q.id = n.query_id\nJOIN semantic_products_text p ON p.id = n.neighbor_id\nORDER BY n.query_id, n.rank\n", "sql_sha256": "6bd4db00ba705b0bcb2bfa6c8c91e465e1e4546148472df571cef28d7ba9d36c", "illustrated_run": "8850d8ec-b120-45b7-adf3-88377fccbaed", "illustrated_source_product_id": 1248764, "illustrated_rows": [ { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 1, "neighbor_id": 1271346, "candidate_title": "Sony WH-1000XM4 Wireless Noise Canceling Overhead Headphones - Black (Renewed)", "cosine_similarity": 0.9512642621994019 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 2, "neighbor_id": 1548411, "candidate_title": "Sony WH-1000XM4 Wireless Industry Leading Noise Cancelling Over-Ear Headphones with Mic for Hands Free Calling and Alexa, Black WH-1000XM4/B Bundle w/Case + Deco Gear Portable Charger + Gym Bag", "cosine_similarity": 0.9046233296394348 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 3, "neighbor_id": 1267625, "candidate_title": "Sony WH-1000XM4 Wireless Noise Canceling Over-Ear Headphones (Silver) Bundle with 10000 mAh Ultra-Portable LED Display Wireless Quick Charge Battery Bank (Black) (2 Items)", "cosine_similarity": 0.8835749626159668 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 4, "neighbor_id": 1585101, "candidate_title": "Sony WH-1000XM4 Wireless Industry Leading Noise Canceling Overhead Headphones with Mic (Black) Bundle with Find My Bluetooth Locator Keychain (2 Items)", "cosine_similarity": 0.8800780177116394 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 5, "neighbor_id": 1344304, "candidate_title": "Sony WH-1000XM4 Wireless Noise Cancelling Over-Ear Headphones with Mic (Silver) in-Ear Wireless Headphones Bundle (2 Items)", "cosine_similarity": 0.8763421773910522 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 6, "neighbor_id": 1593899, "candidate_title": "Sony WH-1000XM4 Wireless Noise Canceling Over-Ear Headphones (Silver) Bundle with Headphone Hanger Mount (2 Items)", "cosine_similarity": 0.8736070990562439 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 7, "neighbor_id": 1297898, "candidate_title": "Sony WH-1000XM4 Wireless Noise Canceling Overhead Headphones with Mic for Phone-Call, Voice Control, Black, with USB Wall Adapter and Microfiber Cleaning Cloth - Bundle", "cosine_similarity": 0.8707782030105591 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 8, "neighbor_id": 1355803, "candidate_title": "Sony WH-1000XM4 Wireless Noise-Canceling Over-Ear Headphones (Black WH1000XM4/B) Bundle + Wall Charger with USB Type-C Cable", "cosine_similarity": 0.8578447699546814 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 9, "neighbor_id": 1367556, "candidate_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones, Blue (Renewed)", "cosine_similarity": 0.8517288565635681 }, { "query_id": 1248764, "source_title": "Sony WH-1000XM4 Wireless Premium Noise Canceling Overhead Headphones with Mic for Phone-Call and Alexa Voice Control, Midnight Blue WH1000XM4", "rank": 10, "neighbor_id": 1448833, "candidate_title": "Sony WH-1000XM4 Wireless Noise Canceling Over-Ear Headphones (Black) WLA-NS7 Wireless TV Adapter Bundle (2 Items)", "cosine_similarity": 0.8406590819358826 } ] }, "headphone_emerging": { "runs": 3, "statuses": [ "ok", "ok", "ok" ], "median_s": 0.7515872349613346, "min_s": 0.7323827510117553, "max_s": 0.9018790269619785, "rows": 10, "observations": [ { "query_id": "f11838e2-3a9a-42c9-b4fd-f9fba8adbd29", "post_native_execution_path": "mixed", "host_datafusion_boundary_count": "20", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "8227210743", "gpu_resident_input_handoffs": "1", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "7", "device_to_host_materialized_bytes": "1990780", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "native_partition_preserving_sort_unsupported" ] }, { "query_id": "eeaa3407-814a-4686-9dcf-f38479caacf4", "post_native_execution_path": "mixed", "host_datafusion_boundary_count": "20", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "8227210743", "gpu_resident_input_handoffs": "1", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "7", "device_to_host_materialized_bytes": "1990780", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "native_partition_preserving_sort_unsupported" ] }, { "query_id": "dd73b315-c205-4d8e-b1a2-f98d8f483090", "post_native_execution_path": "mixed", "host_datafusion_boundary_count": "20", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "8227210743", "gpu_resident_input_handoffs": "1", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "7", "device_to_host_materialized_bytes": "1990780", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "native_partition_preserving_sort_unsupported" ] } ], "sql": "WITH assignments AS (\n SELECT id, cluster_id\n FROM cuvs_kmeans(\n input => (SELECT v.id, v.vector\n FROM semantic_reviews_vectors v\n JOIN semantic_reviews_text r ON r.id = v.id\n WHERE r.category_l3 = 'Headphones & Earbuds'\n ORDER BY v.id),\n n_clusters => 24, n_init => 5, max_iter => 100, tol => 0.0001)\n), issues AS (\n SELECT r.product_id, a.cluster_id,\n SUM(CASE WHEN EXTRACT(YEAR FROM r.review_time) = 2021 THEN 1 ELSE 0 END) AS n2021,\n SUM(CASE WHEN EXTRACT(YEAR FROM r.review_time) = 2022 THEN 1 ELSE 0 END) AS n2022,\n MIN(CASE WHEN EXTRACT(YEAR FROM r.review_time) = 2022 THEN r.id END) AS example_id\n FROM assignments a JOIN semantic_reviews_text r ON r.id = a.id\n GROUP BY r.product_id, a.cluster_id\n), totals AS (\n SELECT *, SUM(n2021) OVER (PARTITION BY product_id) AS total2021,\n SUM(n2022) OVER (PARTITION BY product_id) AS total2022\n FROM issues\n)\nSELECT p.product_id, p.title, t.cluster_id,\n t.n2021, t.total2021, t.n2022, t.total2022,\n 100.0 * t.n2021 / t.total2021 AS share2021_pct,\n 100.0 * t.n2022 / t.total2022 AS share2022_pct,\n 100.0 * t.n2022 / t.total2022 - 100.0 * t.n2021 / t.total2021 AS change_pp,\n t.example_id, r.embedding_text AS example_review\nFROM totals t\nJOIN products p ON p.product_id = t.product_id\nJOIN semantic_reviews_text r ON r.id = t.example_id\nWHERE t.total2021 >= 30 AND t.total2022 >= 30 AND t.n2022 >= 10\n AND 1.0 * t.n2022 / t.total2022 - 1.0 * t.n2021 / t.total2021 > 0.10\nORDER BY change_pp DESC, p.product_id, t.cluster_id\nLIMIT 10\n", "sql_sha256": "3f52ce07e96032f17366425687e3daa6d866b86d8dc321489190a7cbd844f2f4", "illustrated_run": "f11838e2-3a9a-42c9-b4fd-f9fba8adbd29", "illustrated_rows": [ { "product_id": 1507685, "title": "Altec Lansing NanoPods - Truly Wireless Earbuds with Charging Case, TWS Waterproof Bluetooth Earbuds with Touch Controls for Travel, Sports, Running, Working (Royal Blue)", "cluster_id": 7, "n2021": 1, "total2021": 42, "n2022": 10, "total2022": 33, "share2021_pct": 2.380952380952381, "share2022_pct": 30.303030303030305, "change_pp": 27.922077922077925, "example_id": 14707192, "example_review": "Won’t charge This is my second set. The first set was amazing. Second set will not charge. Dissatisfied" }, { "product_id": 1545158, "title": "Jabra Talk 65 Mono - Premium Wireless Single Ear Headset - 2 Built-in Noise Cancelling Microphones, Media Streaming, Up to 100 Meters Bluetooth Range - Black", "cluster_id": 10, "n2021": 4, "total2021": 73, "n2022": 14, "total2022": 58, "share2021_pct": 5.47945205479452, "share2022_pct": 24.137931034482758, "change_pp": 18.658478979688237, "example_id": 14777699, "example_review": "Un mal diseño No tiene control de volumen y se prende pegao a la cara" }, { "product_id": 1564240, "title": "Sony Wireless Headphones WH-CH510: Wireless Bluetooth On-Ear Headset with Mic for phone-call, Blue", "cluster_id": 8, "n2021": 6, "total2021": 67, "n2022": 14, "total2022": 55, "share2021_pct": 8.955223880597014, "share2022_pct": 25.454545454545453, "change_pp": 16.49932157394844, "example_id": 9948566, "example_review": "Worst wireless headphones… beats is better I had beats wireless hand phones which were expensive but hands down, the best. You pay what you get. These are so uncomfortable especially when you wear them for a long time. 30 minutes in and my ears were hurting so I had to take it off. This occurred every time I wore them. Also the Bluetooth is so picky. It says it’s connected but not really. I had to return it. $30 is too much to waste money on headphones you barely use." }, { "product_id": 1341508, "title": "Bose Sport Open Earbuds — True Wireless Open Ear Headphones - Sweat Resistant for Running, Walking and Workouts, Black", "cluster_id": 19, "n2021": 3, "total2021": 31, "n2022": 13, "total2022": 52, "share2021_pct": 9.67741935483871, "share2022_pct": 25.0, "change_pp": 15.32258064516129, "example_id": 785062, "example_review": "Great sound, super uncomfortable Love the rich sounds that come from these for music, they also have really good clarity for calls and were easy to pair with my phone and tablet through the Bose app, however they are super uncomfortable. The part that goes behind your ear is way too bulky and rather painful - it pushed my ears out and squeezed them to a point where my ears turned red and started to throb after just a few minutes of wearing. I feel like I have average size and shaped ears so I'm puzzled why they would design these with such a bulky ear piece to house all the tech. There is also a seam in the plastic that was placed a poor spot and rubs on the back of the ear causing further irritation. I wasn't worried about them falling off while walking though and they might work better for someone with larger ears or ears that naturally stick out a bit further from their head which might allow space for the bulky earpiece but that likely still wouldn't do much for the overall \"pinching\" feeling and the seam irritation. I really wanted these to work as I hate ear buds that go in your ears and block out all the noises around you but sadly after wearing them for just 30 minutes, I knew I was going to have to return these. Perhaps they will come back with a better design in the future that has a flatter, more comfortable wear. very disappointing for the overall price and brand expectations when it came to fit and comfort, on par for sound quality though." }, { "product_id": 1435838, "title": "Samsung Galaxy Buds Plus, True Wireless Earbuds (Wireless Charging Case Included), White – US Version", "cluster_id": 9, "n2021": 53, "total2021": 482, "n2022": 27, "total2022": 104, "share2021_pct": 10.995850622406639, "share2022_pct": 25.96153846153846, "change_pp": 14.96568783913182, "example_id": 767918, "example_review": "Broke after two years I would have said I liked them, they were comfortable and had a great battery life. Unfortunately after a little over 2 years of very light use, one of the buds stopped working. For the cost, I'd expect them to be a little more durable." }, { "product_id": 1377541, "title": "TREBLAB X3-Pro - Wireless Earbuds with Earhooks - 45H Playtime, aptX, IPX7 Waterproof Earphones for Running & Workout - Sport Bluetooth Headphones with Charging case - Built-in Microphone", "cluster_id": 15, "n2021": 2, "total2021": 32, "n2022": 15, "total2022": 71, "share2021_pct": 6.25, "share2022_pct": 21.12676056338028, "change_pp": 14.87676056338028, "example_id": 1446330, "example_review": "Horrible connectivity and Charging Dock pros: sound quality great, noise cancellation great cons: The charging dock is horrendous. What these earphones decide to do is constantly disconnect from the metal leds inside the dock, which causes the buds to connect to your device while inside the dock. This is especially annoying when you are listening to your phone via your car's audio device, and then the buds decide to connect to your phone all of a sudden. other devices do not do this as when they are in the dock the buds stay disconnected. Also trying to get the buds to line up with the leds so that they charge is a struggle. These worked at first for a few months, over time they are unreliable for the above reasons. I have tried cleaning everything and resetting the devices, but they don't work." }, { "product_id": 1569701, "title": "6S Wireless Bluetooth Headphones Over Ear, Hi-Fi Stereo Foldable Wireless Stereo Headsets Earbuds with Built-in Mic, Volume Control, FM for iPhone/Samsung/iPad/PC (White & Gold)", "cluster_id": 6, "n2021": 7, "total2021": 63, "n2022": 34, "total2022": 138, "share2021_pct": 11.11111111111111, "share2022_pct": 24.63768115942029, "change_pp": 13.526570048309178, "example_id": 4552805, "example_review": "DON'T BUY!!! CHEAP. BREAKS EASILY! I guess you get what you pay for! It seems that the headset is poorly made and the set breaks easily.. Just be prepared to buy a new one faster than you thought!" }, { "product_id": 1230203, "title": "Jaybird Vista 2 True Wireless Bluetooth Headphones With Charging Case - Premium Sound, ANC, Sport Fit, 24 Hour Battery, Waterproof Earbuds With Military-Grade Durability - Black", "cluster_id": 9, "n2021": 1, "total2021": 33, "n2022": 17, "total2022": 109, "share2021_pct": 3.0303030303030303, "share2022_pct": 15.596330275229358, "change_pp": 12.566027244926328, "example_id": 14327079, "example_review": "Pauses randomly Look, the Jaybird 1s were a perfect 5 for what I wanted and needed. Sounded great, worked well, stayed in place when running or at the gym. I was pretty excited when I ordered these. They sound about the same, but they have app support which is pretty cool. However, they're nearly useless... They randomly pause the show or music I'm listening to. I know what you're going to say \"Autopause is turned on!\" Nay. I turn it off every time I use them (because seeing a default setting clearly didn't cross anyone's mind). After they pause, they end up out of sync, one bud on, the other off. Only way to fix it is to put both buds back in the charger and reset. This happens every 5-10 minutes minimum, often more frequently. I do it probably twice until I give up and finish my workout without music, defeating the purpose of having them in the first place. Would not buy again." }, { "product_id": 1569701, "title": "6S Wireless Bluetooth Headphones Over Ear, Hi-Fi Stereo Foldable Wireless Stereo Headsets Earbuds with Built-in Mic, Volume Control, FM for iPhone/Samsung/iPad/PC (White & Gold)", "cluster_id": 13, "n2021": 2, "total2021": 63, "n2022": 21, "total2022": 138, "share2021_pct": 3.1746031746031744, "share2022_pct": 15.217391304347826, "change_pp": 12.042788129744652, "example_id": 4492670, "example_review": "Don’t bother Just bought these in December and they are broken already." }, { "product_id": 1587636, "title": "LUDOS Clamor Wired Earbuds in Ear, Noise Isolating Headphones with Microphone, 3.5mm Jack Plug, Mic and Volume Control, Memory Foam, Deep Bass, Tangle-Free Cord - Purple", "cluster_id": 10, "n2021": 9, "total2021": 107, "n2022": 15, "total2022": 76, "share2021_pct": 8.411214953271028, "share2022_pct": 19.736842105263158, "change_pp": 11.32562715199213, "example_id": 3060644, "example_review": "Constant white noise hiss Bummer, was hoping these would work well for zoom meetings with my work issued Dell laptop. They work but I have to listen to constant white noise hiss and others say my voice sounds echo-y. I do not have these issues with my Bluetooth headset, but my company security settings messed up the ability to use that one. Will be looking for a wired headset with decent sound quality. Fun." } ] }, "inline_kmeans_topk": { "runs": 3, "statuses": [ "ok", "ok", "ok" ], "median_s": 0.7763096200069413, "min_s": 0.7651936759939417, "max_s": 0.7992983189760707, "rows": 18, "observations": [ { "query_id": "512d57e0-fd88-4045-a862-429443a70856", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "773154830", "gpu_resident_input_handoffs": "4", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "2", "device_to_host_materialized_bytes": "10129", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "7feff5cb-492d-4693-9af3-55b932a3bdd9", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "773148890", "gpu_resident_input_handoffs": "4", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "2", "device_to_host_materialized_bytes": "10129", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "b29d6a79-39a9-40e3-bc43-6976794e051a", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "773150186", "gpu_resident_input_handoffs": "4", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "2", "device_to_host_materialized_bytes": "10129", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] } ], "sql": "WITH complaint_features AS (\n SELECT\n p.product_id AS id,\n COUNT(*) AS negative_reviews,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%battery%' THEN 1.0 ELSE 0.0 END) AS REAL) AS battery,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%connect%' OR LOWER(r.text) LIKE '%bluetooth%' THEN 1.0 ELSE 0.0 END) AS REAL) AS connection,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%sound%' OR LOWER(r.text) LIKE '%volume%' THEN 1.0 ELSE 0.0 END) AS REAL) AS sound,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%comfort%' OR LOWER(r.text) LIKE '%fit%' THEN 1.0 ELSE 0.0 END) AS REAL) AS comfort,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%broke%' OR LOWER(r.text) LIKE '%stopped working%' THEN 1.0 ELSE 0.0 END) AS REAL) AS durability,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%charg%' THEN 1.0 ELSE 0.0 END) AS REAL) AS charging,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%microphone%' OR LOWER(r.text) LIKE '%mic %' THEN 1.0 ELSE 0.0 END) AS REAL) AS microphone,\n CAST(AVG(CASE WHEN LOWER(r.text) LIKE '%return%' OR LOWER(r.text) LIKE '%refund%' THEN 1.0 ELSE 0.0 END) AS REAL) AS returns\n FROM reviews AS r\n JOIN products AS p\n ON r.product_id = p.product_id\n WHERE p.category_l2 = 'Headphones, Earbuds & Accessories'\n AND r.review_time >= TIMESTAMP '2022-01-01 00:00:00'\n AND r.review_time < TIMESTAMP '2023-01-01 00:00:00'\n AND r.rating <= 2\n AND r.text IS NOT NULL\n GROUP BY p.product_id\n HAVING COUNT(*) >= 10\n),\n\nclusters AS (\n SELECT\n id,\n cluster_id\n FROM cuvs_kmeans(\n input => (\n SELECT\n id,\n COALESCE(battery, CAST(0 AS REAL)) AS battery,\n COALESCE(connection, CAST(0 AS REAL)) AS connection,\n COALESCE(sound, CAST(0 AS REAL)) AS sound,\n COALESCE(comfort, CAST(0 AS REAL)) AS comfort,\n COALESCE(durability, CAST(0 AS REAL)) AS durability,\n COALESCE(charging, CAST(0 AS REAL)) AS charging,\n COALESCE(microphone, CAST(0 AS REAL)) AS microphone,\n COALESCE(returns, CAST(0 AS REAL)) AS returns\n FROM complaint_features\n ORDER BY id\n ),\n n_clusters => 6,\n n_init => 5\n )\n),\n\nprofiles AS (\n SELECT\n CAST(k.cluster_id AS BIGINT) AS id,\n COALESCE(CAST(AVG(c.battery) AS REAL), CAST(0 AS REAL)) AS battery,\n COALESCE(CAST(AVG(c.connection) AS REAL), CAST(0 AS REAL)) AS connection,\n COALESCE(CAST(AVG(c.sound) AS REAL), CAST(0 AS REAL)) AS sound,\n COALESCE(CAST(AVG(c.comfort) AS REAL), CAST(0 AS REAL)) AS comfort,\n COALESCE(CAST(AVG(c.durability) AS REAL), CAST(0 AS REAL)) AS durability,\n COALESCE(CAST(AVG(c.charging) AS REAL), CAST(0 AS REAL)) AS charging,\n COALESCE(CAST(AVG(c.microphone) AS REAL), CAST(0 AS REAL)) AS microphone,\n COALESCE(CAST(AVG(c.returns) AS REAL), CAST(0 AS REAL)) AS returns\n FROM clusters AS k\n JOIN complaint_features AS c\n ON k.id = c.id\n GROUP BY k.cluster_id\n)\n\nSELECT\n t.id AS cluster_id,\n t.rank,\n CASE t.position\n WHEN 0 THEN 'battery'\n WHEN 1 THEN 'connection'\n WHEN 2 THEN 'sound'\n WHEN 3 THEN 'comfort'\n WHEN 4 THEN 'durability'\n WHEN 5 THEN 'charging'\n WHEN 6 THEN 'microphone'\n WHEN 7 THEN 'returns'\n END AS issue,\n t.value AS mean_product_mention_rate\nFROM cuvs_top_k(\n input => (\n SELECT\n id,\n battery,\n connection,\n sound,\n comfort,\n durability,\n charging,\n microphone,\n returns\n FROM profiles\n ),\n k => 3,\n select => 'max'\n) AS t\nORDER BY\n t.id,\n t.rank;\n\n", "sql_sha256": "c062e317073f927f8baaf709397feb6557d87e042b23a9cd14debf49c4f6e779", "illustrated_run": "512d57e0-fd88-4045-a862-429443a70856", "illustrated_rows": [ { "cluster_id": 0, "rank": 1, "issue": "durability", "mean_product_mention_rate": 0.35986992716789246 }, { "cluster_id": 0, "rank": 2, "issue": "sound", "mean_product_mention_rate": 0.18610817193984985 }, { "cluster_id": 0, "rank": 3, "issue": "returns", "mean_product_mention_rate": 0.1122603714466095 }, { "cluster_id": 1, "rank": 1, "issue": "sound", "mean_product_mention_rate": 0.4509941637516022 }, { "cluster_id": 1, "rank": 2, "issue": "charging", "mean_product_mention_rate": 0.24197699129581451 }, { "cluster_id": 1, "rank": 3, "issue": "connection", "mean_product_mention_rate": 0.24094454944133759 }, { "cluster_id": 2, "rank": 1, "issue": "charging", "mean_product_mention_rate": 0.39991021156311035 }, { "cluster_id": 2, "rank": 2, "issue": "sound", "mean_product_mention_rate": 0.2141246348619461 }, { "cluster_id": 2, "rank": 3, "issue": "connection", "mean_product_mention_rate": 0.1516951024532318 }, { "cluster_id": 3, "rank": 1, "issue": "sound", "mean_product_mention_rate": 0.48470911383628845 }, { "cluster_id": 3, "rank": 2, "issue": "comfort", "mean_product_mention_rate": 0.1729457527399063 }, { "cluster_id": 3, "rank": 3, "issue": "returns", "mean_product_mention_rate": 0.14595681428909302 }, { "cluster_id": 4, "rank": 1, "issue": "comfort", "mean_product_mention_rate": 0.24262027442455292 }, { "cluster_id": 4, "rank": 2, "issue": "returns", "mean_product_mention_rate": 0.08225435018539429 }, { "cluster_id": 4, "rank": 3, "issue": "durability", "mean_product_mention_rate": 0.05911022424697876 }, { "cluster_id": 5, "rank": 1, "issue": "sound", "mean_product_mention_rate": 0.26485252380371094 }, { "cluster_id": 5, "rank": 2, "issue": "connection", "mean_product_mention_rate": 0.19398491084575653 }, { "cluster_id": 5, "rank": 3, "issue": "returns", "mean_product_mention_rate": 0.11981520056724548 } ] }, "headphone_umap2": { "runs": 3, "statuses": [ "ok", "ok", "ok" ], "median_s": 3.8821365920011885, "min_s": 3.634793533012271, "max_s": 4.040503346011974, "rows": 9967, "observations": [ { "query_id": "2f1b7c06-878f-4034-bbc4-b9a6605396ed", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "1888152609", "gpu_resident_input_handoffs": "2", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "4", "device_to_host_materialized_bytes": "13745306", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "47b502f1-9e37-4ad9-8b58-e419b2cf4c94", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "1888152609", "gpu_resident_input_handoffs": "2", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "4", "device_to_host_materialized_bytes": "13745306", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] }, { "query_id": "95410fdc-cd78-4d05-816a-2a5e0eb44c15", "post_native_execution_path": "native", "host_datafusion_boundary_count": "0", "device_ordinal": "0", "query_grant_bytes": "34359738368", "max_task_observed_peak_bytes": "1888152609", "gpu_resident_input_handoffs": "2", "host_to_device_uploaded_bytes": "0", "device_to_host_materializations": "4", "device_to_host_materialized_bytes": "13745306", "allocator_high_water_bytes": "none", "planning_reason_codes": [ "consumer_producer_binding_required" ] } ], "sql": "SELECT u.id, u.dim_0, u.dim_1, r.product_id,\n p.title AS product_title, EXTRACT(YEAR FROM r.review_time) AS review_year,\n r.rating, r.title, r.embedding_text\nFROM cuml_umap(\n input => (SELECT v.id, v.vector\n FROM semantic_reviews_vectors v\n JOIN semantic_reviews_text r ON r.id = v.id\n WHERE r.category_l3 = 'Headphones & Earbuds' AND v.sample_rank < 100000\n ORDER BY v.id),\n n_components => 2, n_neighbors => 30, min_dist => 0.1, random_state => 42) u\nJOIN semantic_reviews_text r ON r.id = u.id\nJOIN products p ON p.product_id = r.product_id\nORDER BY u.id\n", "sql_sha256": "9e2f041db29501d216c2c0a08a6ab6c0ef9693461a80046c8d19eb1e9a842a85", "illustrated_run": "2f1b7c06-878f-4034-bbc4-b9a6605396ed", "website_sample": { "path": "website/src/data/amazonReviewMap.json", "query_id": "2f1b7c06-878f-4034-bbc4-b9a6605396ed", "sql": "SELECT u.id, u.dim_0, u.dim_1, r.product_id,\n p.title AS product_title, EXTRACT(YEAR FROM r.review_time) AS review_year,\n r.rating, r.title, r.embedding_text\nFROM cuml_umap(\n input => (SELECT v.id, v.vector\n FROM semantic_reviews_vectors v\n JOIN semantic_reviews_text r ON r.id = v.id\n WHERE r.category_l3 = 'Headphones & Earbuds' AND v.sample_rank < 100000\n ORDER BY v.id),\n n_components => 2, n_neighbors => 30, min_dist => 0.1, random_state => 42) u\nJOIN semantic_reviews_text r ON r.id = u.id\nJOIN products p ON p.product_id = r.product_id\nORDER BY u.id\n", "displayed_rows": 1200, "fitted_rows": 9967, "sampling": "evenly spaced positions after ORDER BY id", "sha256": "0c8421b9505b18bc5440d9f42b4844fcbd2ae53aca7be218b2b3b29f3a6404ac" } } } }