Skip to content

brazilianEcommerce.md

Latest commit

 

History

History
507 lines (422 loc) · 46.9 KB

brazilianEcommerce.md

File metadata and controls

507 lines (422 loc) · 46.9 KB

Brazilian E-Commerce

  • URL: https://www.kaggle.com/datasets/olistbr/brazilian-ecommerce
  • Format: CSV
  • Description: This is a Brazilian ecommerce public dataset of orders made at Olist Store. The dataset has information of 100k orders from 2016 to 2018 made at multiple marketplaces in Brazil. Its features allows viewing an order from multiple dimensions: from order status, price, payment and freight performance to customer location, product attributes and finally reviews written by customers. We also released a geolocation dataset that relates Brazilian zip codes to lat/lng coordinates.

Datasets

Fields cardinality processed_olist_customers_dataset_20K.csv

  • (0,0) No relationship between values in the row and column fields
  • (0,1) One value in the row field is related to at most one value in the column field
  • (0,2) One value in the row field can be related to none, one or several values in the column field
  • (1,1) One value in the row filed is only related to one value in the column field
  • (1,2) One value in the row field can be related to one or several values in the column field
  • Inconsistencies:
    • A value of a "customer_city" should only be associated with a value of a "customer_state"

Ontologies

Mapping of the classes, their relationships and attributes between the different ontologies, as well as the Hits@1 metric obtained in the alignment experiments for the AttrE, BootEA, Aligne, GCN_Align and SEA methods.

Class mapping

Mapping

Mapping Dataset RDF attr_triples rel_triples
Basic processed_olist_customers_dataset_20K.csv 120000 triples 100000 triples 20000 triples
Gold processed_olist_customers_dataset_20K.csv 152758 triples 50857 triples 101901 triples
LLM processed_olist_customers_dataset_20K.csv 84357 triples 20000 triples 64357 triples
Transactions processed_olist_customers_dataset_20K.csv 39686 triples 19843 triples 19843 triples

Gold - Gold

High level graph generated for the Gold ontology. Nodes of the same color represent the instance, class type and attributes of the same class. The instance node indicates the number of entities in the test set. Blue arrows indicate attributes and black arrows indicate relationships. The cardinality is determined for each arrow. If it is not (1,1), the probability that it was (1,1) is indicated. The names in parentheses indicate the referenced column in the CSV source file.

Graph Gold

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es/ontology/bre/CustomerIdOrder https://vocab.um.es/ontology/bre/CustomerIdOrder 20000 (39.33%) 8015 (39.40%)
https://vocab.um.es/ontology/bre/CustomerAccount https://vocab.um.es/ontology/bre/CustomerAccount 19843 (39.02%) 7903 (38.85%)
https://vocab.um.es/ontology/bre/State https://vocab.um.es/ontology/bre/State 27 (0.05%) 12 (0.06%)
https://vocab.um.es/ontology/bre/City https://vocab.um.es/ontology/bre/City 2247 (4.42%) 900 (4.42%)
https://vocab.um.es/ontology/bre/ZipCodePrefix https://vocab.um.es/ontology/bre/ZipCodePrefix 8740 (17.18%) 3513 (17.27%)
Total related entities 50857 20343
Split Training (50%) Test (40%) Validation (10%)
1 25428 20343 5086

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerIdOrder (#/%) CustomerAccount (#/%) ZipCodePrefix (#/%) City (#/%) State (#/%)
AlignE 1 20343 71.69 86.29 88.40 151 0.78 45471 5619 (70.11%) 5069 (64.14%) 3083 (87.76%) 800 (88.89%) 12 (100%)
AliNet 1 20343 85.27 95.33 96.10 14 0.90 166191 - - - - -
AttrE 1 20343 96.23 99.01 99.54 1 0.97 59878 7731 (96.46%) 7580 (95.91%) 3382 (96.27%) 871 (96.78%) 12 (100%)
BootEA 1 20343 85.46 95.05 96.33 54 0.90 104864 6709 (83.71%) 6586 (83.34%) 3233 (92.03%) 845 (93.89%) 12 (100%)
BootEA_RotatE 1 20343 81.33 95.66 96.95 10 0.88 146977 - - - - -
BootEA_TransH 1 20343 E r r o r
Conve 1 20343 E r r o r
GCN_Align 1 20343 79.97 90.28 91.39 124 0.85 15364 6394 (79.78%) 5893 (74.57%) 3135 (89.24%) 834 (92.67%) 12 (100%)
GMNN 1 20343 E r r o r
HolE 1 20343 63.89 73.10 75.70 489 0.68 179720 - - - - -
IMUSE 1 20343 95.07 98.85 99.10 2 0.97 11628 - - - - -
IPTransE 1 20343 E r r o r
JAPE 1 20343 66.27 75.95 77.88 523 0.70 19948 - - - - -
KDCoE 1 20343 E r r o r
MTransE 1 20343 61.34 68.10 69.75 695 0.64 17384 - - - - -
MultiKE 1 20343 E r r o r
ProjE 1 20343 31.14 43.10 45.98 1881 0.36 86923 - - - - -
RDGCN 1 20343 E r r o r
RotatE 1 20343 73.81 93.71 95.80 25 0.82 46838 - - - - -
RSN4EA 1 20343 70.77 80.38 82.05 535 0.75 155310 - - - - -
SEA 1 20343 74.49 92.25 94.07 59 0.82 18675 5650 (70.49%) 5571 (70.49%) 3112 (88.59%) 808 (89.78%) 12 (100%)
SimplE 1 20343 63.56 80.31 83.12 247 0.71 11721 - - - - -
TransD 1 20343 60.58 65.66 66.73 727 0.63 18137 - - - - -
TransH 1 20343 69.11 80.34 82.36 303 0.74 43002 - - - - -
TransR 1 20343 0.07 0.28 0.41 6546 0.00 15842 - - - - -

Basic - Basic

High level graph generated for the Gold ontology. Nodes of the same color represent the instance, class type and attributes of the same class. The instance node indicates the number of entities in the test set. Blue arrows indicate attributes and black arrows indicate relationships. The cardinality is determined for each arrow. If it is not (1,1), the probability that it was (1,1) is indicated. The names in parentheses indicate the referenced column in the CSV source file.

Graph Basic

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es/ontology/bre/CustomerOrder https://vocab.um.es/ontology/bre/CustomerOrder 20000 (100%) 8000 (100%)
Total related entities 20000 8000
Split Training (50%) Test (40%) Validation (10%)
1 10000 8000 2000

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerIdOrder (#/%)
AlignE 1 8000 0.04 0.09 0.11 4001 0.00 2186 3 (0.04%)
AliNet 1 8000 E r r o r
AttrE 1 8000 99.55 99.99 99.99 1 1.00 4950 7964 (99.55)
BootEA 1 8000 0.04 0.06 0.13 4017 0.00 3780 3 (0.04%)
BootEA_RotatE 1 8000 0.01 0.08 0.14 4003 0.00 6949 -
BootEA_TransH 1 8000 E r r o r
Conve 1 8000 E r r o r
GCN_Align 1 8000 0.00 0.04 0.06 3994 0.00 300 0 (0.00%)
GMNN 1 8000 100 100 100 1 1.00 69502 -
HolE 1 8000 0.04 0.09 0.15 3989 0.00 1852 -
IMUSE 1 8000 49.01 49.03 49.04 2075 0.49 1076 -
IPTransE 1 8000 E r r o r
JAPE 1 8000 0.03 0.05 0.11 4010 0.00 871 -
KDCoE 1 8000 E r r o r
MTransE 1 8000 0.01 0.06 0.11 4007 0.00 305 -
MultiKE 1 8000 E r r o r
ProjE 1 8000 0.01 0.04 0.09 4023 0.00 1525 -
RDGCN 1 8000 55.36 55.41 55.48 799 0.55 9609 -
RotatE 1 8000 0.00 0.04 0.14 3993 0.00 2249 -
RSN4EA 1 8000 E r r o r
SEA 1 8000 0.03 0.05 0.15 3967 0.00 156 2 (0.03%)
SimplE 1 8000 0.00 0.05 0.13 4007 0.00 188 -
TransD 1 8000 0.00 0.06 0.10 3975 0.00 319 -
TransH 1 8000 0.00 0.04 0.06 3923 0.00 289 -
TransR 1 8000 0.00 0.03 0.11 4035 0.00 790 -

LLM - LLM

High level graph generated for the Gold ontology. Nodes of the same color represent the instance, class type and attributes of the same class. The instance node indicates the number of entities in the test set. Blue arrows indicate attributes and black arrows indicate relationships. The cardinality is determined for each arrow. If it is not (1,1), the probability that it was (1,1) is indicated. The names in parentheses indicate the referenced column in the CSV source file.

Graph LLM

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es#Customer https://vocab.um.es#Customer 20000 (64.49%) 8021 (64.66%)
https://vocab.um.es#ZipCodePrefix https://vocab.um.es#ZipCodePrefix 8740 (28.18%) 3502 (28.23%)
https://vocab.um.es#City https://vocab.um.es#City 2247 (7.24%) 871 (7.02%)
https://vocab.um.es#State https://vocab.um.es#State 27 (0.09%) 11 (0.09%)
Total related entities 31014 12405
Split Training (50%) Test (40%) Validation (10%)
1 15507 12405 3102

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time Customer (#/%) ZipCodePrefix (#/%) City (#/%) State (#/%)
AlignE 1 12405 60.55 83.20 86.77 130 0.71 13869 4125 (51.43%) 2614 (74.64%) 761 (87.37%) 11 (100%)
AliNet 1 12405 62.96 86.17 88.48 196 0.73 23199 - - - -
AttrE 1 12405 87.55 92.79 94.41 8 0.90 11215 7257 (90.48%) 2832 (80.87%) 760 (87.26%) 11 (100%)
BootEA 1 12405 62.33 85.74 89.20 104 0.73 16214 4249 (52.97%) 2703 (77.18%) 769 (88.29%) 11 (100%)
BootEA_RotatE 1 12405 62.28 87.35 90.79 22 0.73 35216 - - - -
BootEA_TransH 1 12405 E r r o r
Conve 1 12405 E r r o r
GCN_Align 1 12405 60.94 82.90 84.96 105 0.70 588 4111 (51.25%) 2672 (76.30%) 766 (87.94%) 11 (100%)
GMNN 1 12405 E r r o r
HolE 1 12405 48.26 66.72 70.65 387 0.57 12489 - - - -
IMUSE 1 12405 99.61 99.96 99.98 1 1.00 948 - - - -
IPTransE 1 12405 E r r o r
JAPE 1 12405 E r r o r
KDCoE 1 12405 E r r o r
MTransE 1 12405 48.59 65.19 67.22 750 0.56 770 - - - -
MultiKE 1 12405 E r r o r
ProjE 1 12405 23.33 23.52 23.77 3623 0.24 2721 - - - -
RDGCN 1 12405 27.27 31.67 33.94 2389 0.30 16257 - - - -
RotatE 1 12405 60.20 84.51 88.88 30 0.71 4564 - - - -
RSN4EA 1 12405 59.86 82.60 85.01 282 0.70 18706 - - - -
SEA 1 12405 60.86 84.20 88.26 61 0.71 2961 4192 (52.26%) 2592 (74.01%) 755 (86.68%) 11 (100%)
SimplE 1 12405 53.97 76.03 78.57 119 0.64 1949 - - - -
TransD 1 12405 48.26 63.64 66.73 574 0.55 4989 - - - -
TransH 1 12405 58.14 78.89 81.36 292 0.67 6148 - - - -
TransR 1 12405 0.11 0.28 0.56 4223 0.00 3887 - - - -

Transactions - Transactions

High level graph generated for the Gold ontology. Nodes of the same color represent the instance, class type and attributes of the same class. The instance node indicates the number of entities in the test set. Blue arrows indicate attributes and black arrows indicate relationships. The cardinality is determined for each arrow. If it is not (1,1), the probability that it was (1,1) is indicated. The names in parentheses indicate the referenced column in the CSV source file.

Graph Transactions

Input

ent_links Related Entities Split 1 Test
https://purl.org/ontologies/BPA/CustomerAccount https://purl.org/ontologies/BPA/CustomerAccount 19843 (100%) 7937 (100%)
Total related entities 19843 7937
Split Training (50%) Test (40%) Validation (10%)
1 9921 7937 1985

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerAccount (#/%)
AlignE 1 7937 0.01 0.09 0.16 3985 0.00 1890 1 (0.01%)
AliNet 1 7937 E r r o r
AttrE 1 7937 14.57 39.75 55.44 22 0.27 5339 1156 (14.57%)
BootEA 1 7937 0.01 0.06 0.11 3950 0.00 2704 1 (0.01%)
BootEA_RotatE 1 7937 0.01 0.06 0.14 3973 0.00 6975 -
BootEA_TransH 1 7937 E r r o r
Conve 1 7937 E r r o r
GCN_Align 1 7937 0.01 0.04 0.13 3987 0.0 90 1 (0.01%)
GMNN 1 7937 100 100 100 1 1.00 23309 -
HolE 1 7937 0.01 0.10 0.16 3987 0.00 1320 -
IMUSE 1 7937 100 100 100 1 1.00 224 -
IPTransE 1 7937 E r r o r
JAPE 1 7937 E r r o r
KDCoE 1 7937 E r r o r
MTransE 1 7937 0.03 0.05 0.13 3944 0.00 170 -
MultiKE 1 7937 E r r o r
ProjE 1 7937 0.01 0.05 0.05 3950 0.00 1440 -
RDGCN 1 7937 0.66 0.71 0.77 3918 0.01 6612 -
RotatE 1 7937 0.00 0.03 0.04 3967 0.00 1705 -
RSN4EA 1 7937 E r r o r
SEA 1 7937 0.00 0.13 0.19 3944 0.00 179 0 (0.0%)
SimplE 1 7937 0.00 0.05 0.09 3971 0.00 122 -
TransD 1 7937 0.00 0.10 0.19 3982 0.00 276 -
TransH 1 7937 0.00 0.13 0.20 3994 0.00 244 -
TransR 1 7937 0.01 0.05 0.08 3978 0.00 747 -

Gold - Basic

Graph Gold-Basic

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es/ontology/bre/CustomerIdOrder https://vocab.um.es/ontology/bre/CustomerOrder 20000 (100%) 8000 (100%)
Total related entities 20000 8000
Split Training (50%) Test (40%) Validation (10%)
1 10000 8000 2000

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerIdOrder (#/%)
AlignE 1 8000 0.01 0.05 0.11 4007 0.00 9023 1 (0.01%)
AliNet 1 8000 E r r o r
AttrE 1 8000 82.84 97.88 99.15 1 0.89 20589 6627 (82.84%)
BootEA 1 8000 0.03 0.13 0.21 3947 0.00 9179 2 (0.03%)
BootEA_RotatE 1 8000 0.00 0.05 0.10 3993 0.00 21939 -
BootEA_TransH 1 8000 E r r o r
Conve 1 8000 E r r o r
GCN_Align 1 8000 0.01 0.06 0.10 3993 0.00 1051 1 (0.01)
GMNN 1 8000 E r r o r
HolE 1 8000 0.00 0.06 0.16 4017 0.00 12261 -
IMUSE 1 8000 95.86 96.65 96.90 185 0.96 2383 -
IPTransE 1 8000 E r r o r
JAPE 1 8000 0.01 0.09 0.10 3956 0.00 2805 -
KDCoE 1 8000 E r r o r
MTransE 1 8000 0.03 0.06 0.11 4004 0.00 1896 -
MultiKE 1 8000 E r r o r
ProjE 1 8000 0.00 0.04 0.10 4026 0.00 12037 -
RDGCN 1 8000 55.78 55.81 55.88 790 0.56 20120 -
RotatE 1 8000 0.01 0.05 0.10 4009 0.00 11038 -
RSN4EA 1 8000 E r r o r
SEA 1 8000 0.01 0.09 0.18 3982 0.00 679 1 (0.01%)
SimplE 1 8000 0.01 0.08 0.16 3983 0.00 521 -
TransD 1 8000 0.03 0.08 0.15 4009 0.00 1761 -
TransH 1 8000 0.04 0.09 0.11 3978 0.00 1382 -
TransR 1 8000 0.01 0.10 0.15 3969 0.00 4884 -

Gold - LLM

Graph Gold-LLM

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es/ontology/bre/CustomerIdOrder - https://vocab.um.es#Customer 20000 (64.49%) 7991 (64.42%)
https://vocab.um.es/ontology/bre/ZipCodePrefix - https://vocab.um.es#ZipCodePrefix 8740 (28.18%) 3528 (28.44%)
https://vocab.um.es/ontology/bre/City - https://vocab.um.es#City 2247 (7.24%) 875 (7.05%)
https://vocab.um.es/ontology/bre/State - https://vocab.um.es#State 27 (0.09%) 11 (0.09%)
Total related entities 31014 12405
Split Training (50%) Test (40%) Validation (10%)
1 15507 12405 3102

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerIdOrder (#/%) ZipCodePrefix (#/%) City (#/%) State (#/%)
AlignE 1 12405 61.00 83.86 87.06 152 0.71 20913 4091 (51.20%) 2708 (76.76%) 757 (86.51%) 11 (100%)
AliNet 1 12405 62.61 85.26 87.36 279 0.72 24381 - - - -
AttrE 1 12405 72.54 82.47 83.77 82 0.77 25731 5633 (70.49%) 2631 (74.57%) 723 (82.63%) 11 (100%)
BootEA 1 12405 62.93 86.56 89.94 102 0.73 31157 4249 (53.17%) 2779 (78.77%) 768 (87.77%) 11 (100%)
BootEA_RotatE 1 12405 62.48 87.05 90.62 24 0.73 37592 - - - -
BootEA_TransH 1 12405 E r r o r
Conve 1 12405 E r r o r
GCN_Align 1 12405 60.25 81.81 83.99 326 0.70 2570 4001 (50.07%) 2694 (76.36%) 768 (87.77%) 11 (100%)
GMNN 1 12405 E r r o r
HolE 1 12405 40.70 59.28 63.52 572 0.49 41036 - - - -
IMUSE 1 12405 E r r o r
IPTransE 1 12405 E r r o r
JAPE 1 12405 E r r o r
KDCoE 1 12405 E r r o r
MTransE 1 12405 47.57 62.92 64.94 821 0.54 4847 - - - -
MultiKE 1 12405 E r r o r
ProjE 1 12405 0.24 1.02 1.87 4215 0.01 22175 - - - -
RDGCN 1 12405 48.14 55.24 56.80 1223 0.51 42698 - - - -
RotatE 1 12405 61.15 85.75 89.63 34 0.72 20979 - - - -
RSN4EA 1 12405 59.65 82.35 85.18 240 0.70 41051 - - - -
SEA 1 12405 61.17 84.41 88.29 88 0.71 6376 4164 (52.11%) 2669 (75.65%) 744 (85.03%) 11 (100%)
SimplE 1 12405 48.65 69.18 72.08 494 0.58 5598 - - - -
TransD 1 12405 48.95 64.53 67.54 531 0.56 10902 - - - -
TransH 1 12405 43.60 56.07 57.86 843 0.49 4975 - - - -
TransR 1 12405 0.07 0.32 0.62 4071 0.00 8490 - - - -

Gold - Transactions

Graph Gold-Transactions

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es/ontology/bre/CustomerAccount - https://purl.org/ontologies/BPA/CustomerAccount 19843 (100%) 7937 (100%)
Total related entities 19843 7937
Split Training (50%) Test (40%) Validation (10%)
1 9921 7937 1985

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerAccount (#/%)
AlignE 1 7937 0.00 0.08 0.09 3955 0.00 4833 0 (0.0%)
AliNet 1 7937 E r r o r
AttrE 1 7937 3.89 13.03 21.09 125 0.10 4965 309 (3.89)
BootEA 1 7937 0.01 0.04 0.11 3981 0.00 5258 1 (0.01%)
BootEA_RotatE 1 7937 0.00 0.04 0.10 3995 0.00 12665 -
BootEA_TransH 1 7937 E r r o r
Conve 1 7937 E r r o r
GCN_Align 1 7937 0.03 0.09 0.14 3983 0.00 374 2 (0.03%)
GMNN 1 7937 100 100 100 1 1.00 36384 -
HolE 1 7937 0.01 0.03 0.08 3935 0.00 4755 -
IMUSE 1 7937 100 100 100 1 1.00 684 -
IPTransE 1 7937 E r r o r
JAPE 1 7937 E r r o r
KDCoE 1 7937 E r r o r
MTransE 1 7937 0.03 0.06 0.18 4002 0.00 882 -
MultiKE 1 7937 E r r o r
ProjE 1 7937 0.05 0.10 0.16 3971 0.00 5720 -
RDGCN 1 7937 0.69 0.76 0.84 3961 0.01 13089 -
RotatE 1 7937 0.01 0.10 0.19 3973 0.00 5468 -
RSN4EA 1 7937 E r r o r
SEA 1 7937 0.00 0.03 0.10 3936 0.00 498 0 (0.00%)
SimplE 1 7937 0.01 0.04 0.06 3976 0.00 370 -
TransD 1 7937 0.01 0.06 0.11 3990 0.00 1058 -
TransH 1 7937 0.01 0.10 0.18 3953 0.00 925 -
TransR 1 7937 0.03 0.08 0.11 3987 0.00 3285 -

Basic - LLM

Graph Basic-LLM

Input

ent_links Related Entities Split 1 Test
https://vocab.um.es/ontology/bre/CustomerOrder https://vocab.um.es#Customer 20000 (100%) 8000 (100%)
Total related entities 20000 8000
Split Training (50%) Test (40%) Validation (10%)
1 10000 8000 2000

Output

Approach Split Aligned Ent H@1 (%) H@5 (%) H@10 (%) MR MRR [0,1] Run time CustomerOrder (#/%)
AlignE 1 8000 0.00 0.06 0.11 3950 0.00 3438 0 (0.00%)
AliNet 1 8000 E r r o r
AttrE 1 8000 14.91 32.96 43.11 96 0.24 6869 1193 (14.91%)
BootEA 1 8000 0.01 0.05 0.09 4048 0.00 3521 1 (0.01%)
BootEA_RotatE 1 8000 0.00 0.09 0.15 3989 0.00 8743 -
BootEA_TransH 1 8000 E r r o r
Conve 1 8000 E r r o r
GCN_Align 1 8000 0.00 0.11 0.19 4012 0.00 126 0 (0.00%)
GMNN 1 8000 100 100 100 1 1.00 45951 -
HolE 1 8000 0.01 0.08 0.18 3986 0.00 2762 -
IMUSE 1 8000 E r r o r
IPTransE 1 8000 E r r o r
JAPE 1 8000 E r r o r
KDCoE 1 8000 E r r o r
MTransE 1 8000 0.00 0.06 0.09 3969 0.00 366 -
MultiKE 1 8000 E r r o r
ProjE 1 8000 0.03 0.09 0.15 4028 0.00 2001 -
RDGCN 1 8000 56.04 56.06 56.16 774 0.56 6698 -
RotatE 1 8000 0.01 0.06 0.14 3990 0.00 2647 -
RSN4EA 1 8000 E r r o r
SEA 1 8000 0.04 0.15 0.19 3975 0.00 541 3 (0.04%)
SimplE 1 8000 0.00 0.05 0.14 4005 0.00 346 -
TransD 1 8000 0.01 0.06 0.15 4001 0.00 845 -
TransH 1 8000 0.04 0.06 0.13 3946 0.00 704 -
TransR 1 8000 0.01 0.06 0.08 4036 0.00 2282 -

Results

Figure Methods Boxplot. Boxplot of the OpenEA methods according to the Hits@1 metric of the 8 pairwise comparisons of the KGs generated from the Brazilian E-Commerce dataset. The orange solid line represents the median and the red diamond the mean of the experiments that did not lead to an error. The number of experiments without error is shown in brackets next to the name of the method. BootEA-TransH, ConvE, KDCoE and MultiKE are not shown because their error rate is equal to 1 for all datasets.

Modules Boxplot Brazilian E-Commerce

Figure Methods Clustering. Hierarchical clustering (Ward algorithm) of 25 OpenEA methods according to the 5 metrics for the 8 pairwise comparisons of the KGs generated from the Brazilian E-Commerce dataset.

Modules Clusters Brazilian E-Commerce

Figure Experiments Boxplot. Boxplot of the 8 types of pairwise comparison experiments of KGs based on the Hits@1 score obtained by the 25 methods carried out in the Brazilian E-Commerce dataset. The orange solid line represents the median and the red diamond the mean of the experiments that did not lead to an error. The number of modules that did not generate an error is shown in brackets under the name of the experiment represented.

Datasets Boxplot Brazilian E-Commerce

Figure Experiments Clustering. Hierarchical clustering, with method Ward, of the 8 types of pairwise comparison experiments of KG, according to the 5 metrics, with all the values scaled to the range [0,1], of the 25 (at most) OpenEA modules carried out in the Bigbasket Products dataset.

Datasets Clusters Brazilian E-Commerce

Merged methods

Pair Approach CustomerIdOrder (#/%) CustomerAccount (#/%) ZipCodePrefix (#/%) City (#/%) State (#/%)
Basic-Basic AttrE 7964 (99.55%) - - - -
Basic-Basic BootEA 3 (0.04%) - - - -
Basic-Basic AlignE 3 (0.04%) - - - -
Basic-Basic SEA 2 (0.03%) - - - -
Basic-Basic AttrE-BootEA 7964 (99.55%) - - - -
Basic-Basic AttrE-AlignE 7964 (99.55%) - - - -
Basic-Basic AttrE-SEA 7964 (99.55%) - - - -
Basic-Gold AttrE 6627 (82.84%) - - - -
Basic-Gold BootEA 2 (0.03%) - - - -
Basic-Gold AlignE 1 (0.01%) - - - -
Basic-Gold SEA 1 (0.01%) - - - -
Basic-Gold AttrE-BootEA 6627 (82.84%) - - - -
Basic-Gold AttrE-AlignE 6628 (82.85%) - - - -
Basic-Gold AttrE-SEA 6627 (82.84%) - - - -
Basic-LLM AttrE 1193 (14.91%) - - - -
Basic-LLM BootEA 1 (0.01%) - - - -
Basic-LLM AlignE 0 (0.00%) - - - -
Basic-LLM SEA 3 (0.04%) - - - -
Basic-LLM AttrE-BootEA 1194 (14.93%) - - - -
Basic-LLM AttrE-AlignE 1193 (14.91%) - - - -
Basic-LLM AttrE-SEA 1195 (14.94%) - - - -
Gold-Gold AttrE 7731 (96.46%) 7580 (95.91%) 3382 (96.27%) 871 (96.78%) 12 (100%)
Gold-Gold BootEA 6709 (83.71%) 6586 (83.34%) 3233 (92.03%) 845 (93.89%) 12 (100%)
Gold-Gold AlignE 5619 (70.11%) 5069 (64.14%) 3083 (87.76%) 800 (88.89%) 12 (100%)
Gold-Gold SEA 5650 (70.49%) 5571 (70.49%) 3112 (88.59%) 808 (89.78%) 12 (100%)
Gold-Gold AttrE-BootEA 7812 (97.47%) 7707 (97.52%) 3413 (97.15%) 875 (97.22%) 12 (100%)
Gold-Gold AttrE-AlignE 7735 (96.51%) 7583 (95.95%) 3391 (96.53%) 871 (96.78%) 12 (100%)
Gold-Gold AttrE-SEA 7737 (96.53%) 7626 (96.50%) 3384 (96.33%) 872 (96.89%) 12 (100%)
Gold-LLM AttrE 5633 (70.49%) - 2631 (74.57%) 723 (82.63%) 11 (100%)
Gold-LLM BootEA 4249 (53.17%) - 2779 (78.77%) 768 (87.77%) 11 (100%)
Gold-LLM AlignE 4091 (51.20%) - 2708 (76.76%) 757 (86.51%) 11 (100%)
Gold-LLM SEA 4164 (52.11%) - 2669 (75.65%) 744 (85.03%) 11 (100%)
Gold-LLM AttrE-BootEA 6199 (77.57%) - 2807 (79.56%) 768 (87.77%) 11 (100%)
Gold-LLM AttrE-AlignE 6034 (75.51%) - 2737 (77.58%) 757 (86.51%) 11 (100%)
Gold-LLM AttrE-SEA 6102 (76.36%) - 2714 (76.93%) 747 (85.37%) 11 (100%)
Gold-Transactions AttrE - 309 (3.89%) - - -
Gold-Transactions BootEA - 1 (0.01%) - - -
Gold-Transactions AlignE - 0 (0.0%) - - -
Gold-Transactions SEA - 0 (0.00%) - - -
Gold-Transactions AttrE-BootEA - 310 (3.91%) - - -
Gold-Transactions AttrE-AlignE - 309 (3.89%) - - -
Gold-Transactions AttrE-SEA - 309 (3.89%) - - -
LLM-LLM AttrE 7257 (90.48%) - 2832 (80.87%) 760 (87.26%) 11 (100%)
LLM-LLM BootEA 4249 (52.97%) - 2703 (77.18%) 769 (88.29%) 11 (100%)
LLM-LLM AlignE 4125 (51.43%) - 2614 (74.64%) 761 (87.37%) 11 (100%)
LLM-LLM SEA 4192 (52.26%) - 2592 (74.01%) 755 (86.68%) 11 (100%)
LLM-LLM AttrE-BootEA 7299 (91.00%) - 2915 (83.24%) 776 (89.09%) 11 (100%)
LLM-LLM AttrE-AlignE 7267 (90.60%) - 2862 (81.72%) 770 (88.40%) 11 (100%)
LLM-LLM AttrE-SEA 7275 (90.70%) - 2854 (81.50%) 765 (87.83%) 11 (100%)
Transactions-Transactions AttrE - 1156 (14.57%) - - -
Transactions-Transactions BootEA - 1 (0.01%) - - -
Transactions-Transactions AlignE - 1 (0.01%) - - -
Transactions-Transactions SEA - 0 (0.0%) - - -
Transactions-Transactions AttrE-BootEA - 1157 (14.58%) - - -
Transactions-Transactions AttrE-AlignE - 1157 (14.58%) - - -
Transactions-Transactions AttrE-SEA - 1156 (14.56%) - - -