Hugging Face's logo

Datasets:
relbert
/
t_rex_relational_similarity

Modalities:
Text
Languages:
English
Size:
100K<n<1M
Tags:
Libraries:
Datasets
mlcroissant
License:
Dataset card Viewer Files Community
asahi417 commited on
Commit
bd607ef
•
1 Parent(s): 19f4983

fix readme

Browse files
Files changed (37) hide show
  1. .gitattributes +33 -0
  2. README.md +68 -0
  3. data/filter_unified.min_entity_1_max_predicate_10.train.jsonl +3 -0
  4. data/filter_unified.min_entity_1_max_predicate_10.validation.jsonl +3 -0
  5. data/filter_unified.min_entity_1_max_predicate_100.train.jsonl +3 -0
  6. data/filter_unified.min_entity_1_max_predicate_100.validation.jsonl +3 -0
  7. data/filter_unified.min_entity_1_max_predicate_25.train.jsonl +3 -0
  8. data/filter_unified.min_entity_1_max_predicate_25.validation.jsonl +3 -0
  9. data/filter_unified.min_entity_1_max_predicate_50.train.jsonl +3 -0
  10. data/filter_unified.min_entity_1_max_predicate_50.validation.jsonl +3 -0
  11. data/filter_unified.min_entity_2_max_predicate_10.train.jsonl +3 -0
  12. data/filter_unified.min_entity_2_max_predicate_10.validation.jsonl +3 -0
  13. data/filter_unified.min_entity_2_max_predicate_100.train.jsonl +3 -0
  14. data/filter_unified.min_entity_2_max_predicate_100.validation.jsonl +3 -0
  15. data/filter_unified.min_entity_2_max_predicate_25.train.jsonl +3 -0
  16. data/filter_unified.min_entity_2_max_predicate_25.validation.jsonl +3 -0
  17. data/filter_unified.min_entity_2_max_predicate_50.train.jsonl +3 -0
  18. data/filter_unified.min_entity_2_max_predicate_50.validation.jsonl +3 -0
  19. data/filter_unified.min_entity_3_max_predicate_10.train.jsonl +3 -0
  20. data/filter_unified.min_entity_3_max_predicate_10.validation.jsonl +3 -0
  21. data/filter_unified.min_entity_3_max_predicate_100.train.jsonl +3 -0
  22. data/filter_unified.min_entity_3_max_predicate_100.validation.jsonl +3 -0
  23. data/filter_unified.min_entity_3_max_predicate_25.train.jsonl +3 -0
  24. data/filter_unified.min_entity_3_max_predicate_25.validation.jsonl +3 -0
  25. data/filter_unified.min_entity_3_max_predicate_50.train.jsonl +3 -0
  26. data/filter_unified.min_entity_3_max_predicate_50.validation.jsonl +3 -0
  27. data/filter_unified.min_entity_4_max_predicate_10.train.jsonl +3 -0
  28. data/filter_unified.min_entity_4_max_predicate_10.validation.jsonl +3 -0
  29. data/filter_unified.min_entity_4_max_predicate_100.train.jsonl +3 -0
  30. data/filter_unified.min_entity_4_max_predicate_100.validation.jsonl +3 -0
  31. data/filter_unified.min_entity_4_max_predicate_25.train.jsonl +3 -0
  32. data/filter_unified.min_entity_4_max_predicate_25.validation.jsonl +3 -0
  33. data/filter_unified.min_entity_4_max_predicate_50.train.jsonl +3 -0
  34. data/filter_unified.min_entity_4_max_predicate_50.validation.jsonl +3 -0
  35. data/filter_unified.test.jsonl +3 -0
  36. process.py +64 -0
  37. t_rex_relation_similarity.py +81 -0
.gitattributes CHANGED
@@ -52,3 +52,36 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
52
  *.jpg filter=lfs diff=lfs merge=lfs -text
53
  *.jpeg filter=lfs diff=lfs merge=lfs -text
54
  *.webp filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52
  *.jpg filter=lfs diff=lfs merge=lfs -text
53
  *.jpeg filter=lfs diff=lfs merge=lfs -text
54
  *.webp filter=lfs diff=lfs merge=lfs -text
55
+ data/filter_unified.min_entity_2_max_predicate_10.train.jsonl filter=lfs diff=lfs merge=lfs -text
56
+ data/filter_unified.min_entity_4_max_predicate_10.validation.jsonl filter=lfs diff=lfs merge=lfs -text
57
+ data/filter_unified.min_entity_4_max_predicate_100.train.jsonl filter=lfs diff=lfs merge=lfs -text
58
+ data/filter_unified.min_entity_4_max_predicate_50.train.jsonl filter=lfs diff=lfs merge=lfs -text
59
+ data/filter_unified.min_entity_4_max_predicate_100.validation.jsonl filter=lfs diff=lfs merge=lfs -text
60
+ data/filter_unified.test.jsonl filter=lfs diff=lfs merge=lfs -text
61
+ data/filter_unified.min_entity_1_max_predicate_10.validation.jsonl filter=lfs diff=lfs merge=lfs -text
62
+ data/filter_unified.min_entity_1_max_predicate_100.train.jsonl filter=lfs diff=lfs merge=lfs -text
63
+ data/filter_unified.min_entity_2_max_predicate_50.train.jsonl filter=lfs diff=lfs merge=lfs -text
64
+ data/filter_unified.min_entity_3_max_predicate_25.train.jsonl filter=lfs diff=lfs merge=lfs -text
65
+ data/filter_unified.min_entity_3_max_predicate_50.validation.jsonl filter=lfs diff=lfs merge=lfs -text
66
+ data/filter_unified.min_entity_4_max_predicate_10.train.jsonl filter=lfs diff=lfs merge=lfs -text
67
+ data/filter_unified.min_entity_1_max_predicate_50.train.jsonl filter=lfs diff=lfs merge=lfs -text
68
+ data/filter_unified.min_entity_1_max_predicate_50.validation.jsonl filter=lfs diff=lfs merge=lfs -text
69
+ data/filter_unified.min_entity_3_max_predicate_10.validation.jsonl filter=lfs diff=lfs merge=lfs -text
70
+ data/filter_unified.min_entity_1_max_predicate_25.train.jsonl filter=lfs diff=lfs merge=lfs -text
71
+ data/filter_unified.min_entity_2_max_predicate_10.validation.jsonl filter=lfs diff=lfs merge=lfs -text
72
+ data/filter_unified.min_entity_2_max_predicate_100.validation.jsonl filter=lfs diff=lfs merge=lfs -text
73
+ data/filter_unified.min_entity_2_max_predicate_25.train.jsonl filter=lfs diff=lfs merge=lfs -text
74
+ data/filter_unified.min_entity_2_max_predicate_50.validation.jsonl filter=lfs diff=lfs merge=lfs -text
75
+ data/filter_unified.min_entity_3_max_predicate_50.train.jsonl filter=lfs diff=lfs merge=lfs -text
76
+ data/filter_unified.min_entity_4_max_predicate_25.validation.jsonl filter=lfs diff=lfs merge=lfs -text
77
+ data/filter_unified.min_entity_1_max_predicate_25.validation.jsonl filter=lfs diff=lfs merge=lfs -text
78
+ data/filter_unified.min_entity_2_max_predicate_100.train.jsonl filter=lfs diff=lfs merge=lfs -text
79
+ data/filter_unified.min_entity_3_max_predicate_10.train.jsonl filter=lfs diff=lfs merge=lfs -text
80
+ data/filter_unified.min_entity_4_max_predicate_25.train.jsonl filter=lfs diff=lfs merge=lfs -text
81
+ data/filter_unified.min_entity_3_max_predicate_100.train.jsonl filter=lfs diff=lfs merge=lfs -text
82
+ data/filter_unified.min_entity_3_max_predicate_100.validation.jsonl filter=lfs diff=lfs merge=lfs -text
83
+ data/filter_unified.min_entity_3_max_predicate_25.validation.jsonl filter=lfs diff=lfs merge=lfs -text
84
+ data/filter_unified.min_entity_1_max_predicate_10.train.jsonl filter=lfs diff=lfs merge=lfs -text
85
+ data/filter_unified.min_entity_1_max_predicate_100.validation.jsonl filter=lfs diff=lfs merge=lfs -text
86
+ data/filter_unified.min_entity_2_max_predicate_25.validation.jsonl filter=lfs diff=lfs merge=lfs -text
87
+ data/filter_unified.min_entity_4_max_predicate_50.validation.jsonl filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,68 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ license:
5
+ - other
6
+ multilinguality:
7
+ - monolingual
8
+ pretty_name: t_rex
9
+ ---
10
+
11
+ # Dataset Card for "relbert/t_rex_relation_similarity"
12
+ ## Dataset Description
13
+ - **Repository:** [https://hadyelsahar.github.io/t-rex/](https://hadyelsahar.github.io/t-rex/)
14
+ - **Paper:** [https://aclanthology.org/L18-1544/](https://aclanthology.org/L18-1544/)
15
+ - **Dataset:** T-REX
16
+
17
+ ## Dataset Summary
18
+ This is the clean version of [T-REX](https://aclanthology.org/L18-1544/) converted into relation similarity dataset format.
19
+ The original dataset is [`relbert/t_rex`](https://huggingface.co/datasets/relbert/t_rex).
20
+
21
+ - statistics of the train/validation split
22
+
23
+ | data | num of relation types (train) | average num of positive pairs (train) | average num of negative pairs (train) | num of relation types (validation) | average num of positive pairs (validation) | average num of negative pairs (validation) |
24
+ |:-------------------------------|--------------------------------:|----------------------------------------:|----------------------------------------:|-------------------------------------:|---------------------------------------------:|---------------------------------------------:|
25
+ | min_entity_1_max_predicate_100 | 208 | 34 | 7041 | 133 | 6 | 781 |
26
+ | min_entity_1_max_predicate_50 | 204 | 20 | 4111 | 113 | 4 | 455 |
27
+ | min_entity_1_max_predicate_25 | 202 | 12 | 2346 | 71 | 3 | 259 |
28
+ | min_entity_1_max_predicate_10 | 192 | 6 | 1128 | 25 | 2 | 125 |
29
+ | min_entity_2_max_predicate_100 | 188 | 26 | 4847 | 107 | 5 | 537 |
30
+ | min_entity_2_max_predicate_50 | 184 | 16 | 2986 | 85 | 3 | 331 |
31
+ | min_entity_2_max_predicate_25 | 181 | 9 | 1702 | 51 | 3 | 188 |
32
+ | min_entity_2_max_predicate_10 | 171 | 5 | 853 | 13 | 2 | 94 |
33
+ | min_entity_3_max_predicate_100 | 166 | 22 | 3637 | 82 | 5 | 402 |
34
+ | min_entity_3_max_predicate_50 | 157 | 15 | 2321 | 66 | 3 | 257 |
35
+ | min_entity_3_max_predicate_25 | 156 | 9 | 1381 | 37 | 3 | 152 |
36
+ | min_entity_3_max_predicate_10 | 148 | 4 | 684 | 17 | 2 | 75 |
37
+ | min_entity_4_max_predicate_100 | 150 | 20 | 2975 | 73 | 4 | 329 |
38
+ | min_entity_4_max_predicate_50 | 145 | 14 | 1975 | 56 | 3 | 219 |
39
+ | min_entity_4_max_predicate_25 | 141 | 9 | 1212 | 34 | 3 | 133 |
40
+ | min_entity_4_max_predicate_10 | 128 | 4 | 599 | 14 | 2 | 66 |
41
+
42
+ - statistics of the test split
43
+
44
+ | num of relation types (test) | average num of positive pairs (test) | average num of negative pairs (test) |
45
+ |-------------------------------:|---------------------------------------:|---------------------------------------:|
46
+ | 24 | 5 | 117 |
47
+
48
+
49
+ ## Dataset Structure
50
+ ### Data Instances
51
+ An example looks as follows.
52
+ ```
53
+ {
54
+ "relation_type": "[Airline] has a hub in [Location]",
55
+ "positives": [["Korean Air", "Seoul"], ["Asiana Airlines", "Seoul"], ["Cathay Pacific", "Hong Kong"], ["Dragonair", "Hong Kong"], ["Qantas", "Singapore"], ["Air China", "Beijing"], ["Singapore Airlines", "Singapore"]],
56
+ "negatives": [["joint resolution", "United States Congress"], ["joint resolution", "Congress"], ["Great Seal", "United States"], ["trident", "Ukraine"], ["harp", "Ireland"], ["Plantagenet", "England"], ["Pahonia", "Lithuania"], ["slavery", "American Civil War"], ["main asteroid belt", "Solar System"], ["Colorado Desert", "Sonoran Desert"], ["DNA", "genome"], ["Mars", "Solar System"], ["Manchester United", "red"], ["Kermit", "greenness"], ["Ruby", "red"], ["Liberal Party", "red"], ["Macintosh", "Apple"], ["Apple II", "Apple"], ["Apple III", "Apple"], ["PlayStation 2", "Sony"], ["PlayStation 2", "Sony Computer Entertainment"], ["Beatles", "George Martin"], ["Baku", "Azerbaijan"], ["Accra", "Ghana"], ["Amman", "Jordan"], ["Hannover", "Lower Saxony"], ["Agartala", "Tripura"], ["Makassar", "South Sulawesi"], ["Taiwan", "China"], ["Poland", "United Nations"], ["Poland", "Europe"], ["Poland", "European Union"], ["Poland", "NATO"], ["German invasion", "22 June 1941"], ["Operation Barbarossa", "22 June 1941"], ["Brazil", "Catholic Church"], ["Turkey", "Islam"], ["Afghanistan", "Islam"], ["Iraq", "Islam"], ["Finland", "Evangelical Lutheran Church"], ["England", "Roman Catholic"], ["Congress", "United States"], ["Sejm", "Poland"], ["Diet", "Japan"], ["Majlis", "Iran"], ["Riksdag", "Sweden"], ["Croatian Parliament", "Croatia"], ["Knesset", "Israel"], ["Parliament", "Sri Lanka"], ["Russia", "Soviet Union"], ["Ukrainian SSR", "Soviet Union"], ["Royal Flying Corps", "Royal Air Force"], ["Canadian Army", "Canadian Forces"], ["Belarus", "Russian"], ["Russia", "Russian"], ["Ukraine", "Russian"], ["Kerala", "Malayalam"], ["American", "English"], ["zlib license", "Open Source Initiative"], ["EPL", "Open Source Initiative"], ["GNU General Public License", "Open Source Initiative"], ["Wrigley Field", "Cubs"], ["Wrigley Field", "Chicago Cubs"], ["Yankee Stadium", "Yankees"], ["Passaic River", "Newark Bay"], ["Rocky", "Sylvester Stallone"], ["The Godfather", "Francis Ford Coppola"], ["Citizen Kane", "Orson Welles"], ["She Hate Me", "Spike Lee"], ["Raajneeti", "Prakash Jha"], ["Doctor Who", "Patrick Troughton"], ["Doctor Who", "Tom Baker"], ["Jana Gana Mana", "India"], ["President", "White House"], ["Washington", "Federalist Party"], ["George Washington", "Federalist Party"], ["Joseph Stalin", "Communist Party"], ["Mao Zedong", "Communist Party"], ["Lenin", "Communist Party"], ["Nelson Mandela", "ANC"], ["Putin", "Communist Party"], ["Nehru", "Indian National Congress"], ["Nicolas Sarkozy", "UMP"], ["Andreas Papandreou", "PASOK"], ["Tim Cook", "Apple"], ["Israel", "Isaac"], ["Meg", "Peter"], ["Elizabeth II", "Canada"], ["Victor Emmanuel III", "Italy"], ["Umberto I", "Italy"], ["Victor Emmanuel II", "Italy"], ["Brahms", "pianist"], ["Beethoven", "piano"], ["Nicky Hopkins", "pianist"], ["Mozart", "violin"], ["John Zorn", "saxophonist"], ["McCartney", "piano"], ["Russians", "Russian"], ["The Real McCoys", "CBS"], ["Brookside", "Channel 4"], ["The Real McCoys", "ABC"], ["Windows", "Microsoft"], ["Busan", "Gyeongbu Line"], ["Seoul", "Gyeongbu Line"], ["Springer Mountain", "Appalachian Trail"], ["Doctor Who", "BBC One"], ["central time zone", "Illinois"], ["CT", "Canada"], ["Central Time Zone", "Mexico"], ["Central Time Zone", "United States"], ["CT", "American"], ["CT", "Mexico"], ["CT", "United States"], ["central time zone", "Indiana"], ["Central Time Zone", "American"]]
57
+ }
58
+ ```
59
+
60
+ ## Citation Information
61
+ ```
62
+ @inproceedings{elsahar2018t,
63
+ title={T-rex: A large scale alignment of natural language with knowledge base triples},
64
+ author={Elsahar, Hady and Vougiouklis, Pavlos and Remaci, Arslen and Gravier, Christophe and Hare, Jonathon and Laforest, Frederique and Simperl, Elena},
65
+ booktitle={Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
66
+ year={2018}
67
+ }
68
+ ```
data/filter_unified.min_entity_1_max_predicate_10.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d69eda19134d3444648d603fd033f8080e777e6bb93a451860e362c1a3a4dc7b
3
+ size 6966611
data/filter_unified.min_entity_1_max_predicate_10.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:570159dfac5ac040e602c1eb63d4cf17daf5efcdb27992b37abb05099e5751d1
3
+ size 107890
data/filter_unified.min_entity_1_max_predicate_100.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:540cd4c2a7d0d079a7c8ef6275543b646b83ace16d5828f9fe08107b2185f85e
3
+ size 49099973
data/filter_unified.min_entity_1_max_predicate_100.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cbd4daed0f874854451a131913240d37ce833bf480c89b16b58dadedd9a6db48
3
+ size 3476629
data/filter_unified.min_entity_1_max_predicate_25.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b96c169831191190e7f318dcb326a4ed83af043f12c7ff4b090effa1ce563f8
3
+ size 15443818
data/filter_unified.min_entity_1_max_predicate_25.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:291b98c7d0012f1a287936363485c088b338475a607d5df28a4db40f4dc93bc9
3
+ size 618959
data/filter_unified.min_entity_1_max_predicate_50.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d31f3c671319f80ad3499514c265b25323641ac5b34199e8586351ce695cb04
3
+ size 27551730
data/filter_unified.min_entity_1_max_predicate_50.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:155b118d6c11ed96088bf03ee193eedfb21bc133c3726af8df807fc4617ae8b2
3
+ size 1740209
data/filter_unified.min_entity_2_max_predicate_10.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb7a9989cfe1ce3e42a06687a729660005ef0b75ba324ed24050a01a6844d02e
3
+ size 4637780
data/filter_unified.min_entity_2_max_predicate_10.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77286f3ab1897f75301e399483f0260bcca0542f2331170a5d9d6da32023a0a6
3
+ size 39299
data/filter_unified.min_entity_2_max_predicate_100.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d4bd18df9c5e5d1ac028c10f712f74235d04f2a34e011adc23659bb142b8314
3
+ size 29596755
data/filter_unified.min_entity_2_max_predicate_100.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f9e9791ae6b3738e4cc01b340a582082b3ed3ccc916c7c1353a1f2a5dad2c9e
3
+ size 1894548
data/filter_unified.min_entity_2_max_predicate_25.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd4fd074f40125fc6dac140e798fd121ca42037b970774d6b427c3c6ae56ef4e
3
+ size 9729765
data/filter_unified.min_entity_2_max_predicate_25.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dbec44cc2defe593574ee97ac937c10ae260f4f1cdbead38b10110ba41c46398
3
+ size 319030
data/filter_unified.min_entity_2_max_predicate_50.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:155d936dc59513f942c7a5aaac6699acfcd8d373b716105a6c685e60babe0834
3
+ size 17643311
data/filter_unified.min_entity_2_max_predicate_50.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d105d0927e63c9e88654a03c2efd9941ebdd328ddf2d446241ea22ea83f71575
3
+ size 923790
data/filter_unified.min_entity_3_max_predicate_10.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b330c5f5b2a104f8b9d7ea39e95302ec70a5293c26dd32abf84c77af2ab3db83
3
+ size 3123088
data/filter_unified.min_entity_3_max_predicate_10.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d18d22ff5b68bcb52ffe13a2d4d995e1b2ec8cba2d8cf96730a1d2598bda34c4
3
+ size 40927
data/filter_unified.min_entity_3_max_predicate_100.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27f201fdf42c813da9671fe8966dc5dec18a2b70cc0e21220fced5db0e8ba0f0
3
+ size 19103948
data/filter_unified.min_entity_3_max_predicate_100.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f81062ebf81642a53c3c3a52761d619371c792f8614df5b86b24d9c417a7c1b
3
+ size 1042740
data/filter_unified.min_entity_3_max_predicate_25.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b86738b4ae03058500939ee84ff92dce33dca91a0a6f46a99b906c0fd19bec78
3
+ size 6680533
data/filter_unified.min_entity_3_max_predicate_25.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6402797ed016a17bd5d823ad0d3202f52f90440d115b5c369e954036cb67389
3
+ size 189365
data/filter_unified.min_entity_3_max_predicate_50.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1978ac7895610b69d5ccdfabee0f8be172aebc77357745b30808891e2b7aee6
3
+ size 11441982
data/filter_unified.min_entity_3_max_predicate_50.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d092e7fec5ba2f20c58140bc74ea58bed20fba3fe604986b91db7d5a3d623420
3
+ size 541636
data/filter_unified.min_entity_4_max_predicate_10.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57ad5e9a410b2dc06ed7175b57c8f01b134d6b599ceadf7caa09fa87f1f09da7
3
+ size 2352885
data/filter_unified.min_entity_4_max_predicate_10.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68c7cff1fc7a716e285a52b0101bb6391bb41b153e23aacf8d2dbdc81c50959e
3
+ size 29053
data/filter_unified.min_entity_4_max_predicate_100.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fd7df22453c09dab2cc15c95ac6ab58694e0251eea135e6d0db2fcc994123ad
3
+ size 13942787
data/filter_unified.min_entity_4_max_predicate_100.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11663dafe1a985aac2a298ccd36012674d6fe7929def8a63388a68582b84e78f
3
+ size 766223
data/filter_unified.min_entity_4_max_predicate_25.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:152bae2b052920a0543a9ef6716e3c0d155c7a50c54bd30180cd4b4988d729d8
3
+ size 5254636
data/filter_unified.min_entity_4_max_predicate_25.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7aba74fea472a74de983a266aa271cff7c557aaa0b62685636c07723d55fb3b3
3
+ size 146023
data/filter_unified.min_entity_4_max_predicate_50.train.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:600b2dfdac64163841031105bccf0a5228cc1496d5d083cba9b4f2d4a8db0a96
3
+ size 8912834
data/filter_unified.min_entity_4_max_predicate_50.validation.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7863eaf6f090884a5de471d5b38aa0dfff204b1a172f67dbd55be1c2a1ef5b42
3
+ size 391676
data/filter_unified.test.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:167a9d1309ae5a6507679ded0af46a1306c44dd6678800eb0e49b988d458c55b
3
+ size 88341
process.py ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import json
2
+ import os
3
+ from itertools import product
4
+ from statistics import mean
5
+
6
+ import pandas as pd
7
+ from datasets import load_dataset
8
+
9
+
10
+ def process(name, split, output):
11
+ data = load_dataset("relbert/t_rex", name, split=split)
12
+ df = data.to_pandas()
13
+ df.pop('text')
14
+ df.pop('title')
15
+ df['pairs'] = [[i, j] for i, j in zip(df.pop('subject'), df.pop('object'))]
16
+ rel_sim_data = [{
17
+ "relation_type": pred,
18
+ "positives": g['pairs'].values.tolist(),
19
+ "negatives": df[df.predicate != pred]['pairs'].values.tolist()
20
+ } for pred, g in df.groupby("predicate") if len(g) >= 2]
21
+ with open(output, "w") as f:
22
+ f.write('\n'.join([json.dumps(i) for i in rel_sim_data]))
23
+
24
+
25
+ parameters_min_e_freq = [1, 2, 3, 4]
26
+ parameters_max_p_freq = [100, 50, 25, 10]
27
+ os.makedirs("data", exist_ok=True)
28
+
29
+ for min_e_freq, max_p_freq in product(parameters_min_e_freq, parameters_max_p_freq):
30
+ for s in ['train', 'validation']:
31
+ process(
32
+ name=f"filter_unified.min_entity_{min_e_freq}_max_predicate_{max_p_freq}",
33
+ split=s,
34
+ output=f"data/filter_unified.min_entity_{min_e_freq}_max_predicate_{max_p_freq}.{s}.jsonl")
35
+
36
+ process(
37
+ name=f"filter_unified.min_entity_{min_e_freq}_max_predicate_{max_p_freq}",
38
+ split='test',
39
+ output=f"data/filter_unified.test.jsonl")
40
+
41
+
42
+ stats = []
43
+ for min_e_freq, max_p_freq in product(parameters_min_e_freq, parameters_max_p_freq):
44
+ stats_tmp = {"data": f"filter_unified.min_entity_{min_e_freq}_max_predicate_{max_p_freq}"}
45
+ for s in ['train', 'validation']:
46
+ with open(f"data/filter_unified.min_entity_{min_e_freq}_max_predicate_{max_p_freq}.{s}.jsonl") as f:
47
+ tmp = [json.loads(i) for i in f.read().split('\n') if len(i) > 0]
48
+ stats_tmp[f'num of relation types ({s})'] = len(tmp)
49
+ stats_tmp[f'average num of positive pairs ({s})'] = round(mean([len(i['positives']) for i in tmp]))
50
+ stats_tmp[f'average num of negative pairs ({s})'] = round(mean([len(i['negatives']) for i in tmp]))
51
+ stats.append(stats_tmp)
52
+ df_stats = pd.DataFrame(stats)
53
+ df_stats.index = df_stats.pop('data')
54
+ print(df_stats.to_markdown())
55
+
56
+ stats_tmp = {}
57
+ with open("data/filter_unified.test.jsonl") as f:
58
+ tmp = [json.loads(i) for i in f.read().split('\n') if len(i) > 0]
59
+ stats_tmp[f'num of relation types (test)'] = len(tmp)
60
+ stats_tmp[f'average num of positive pairs (test)'] = round(mean([len(i['positives']) for i in tmp]))
61
+ stats_tmp[f'average num of negative pairs (test)'] = round(mean([len(i['negatives']) for i in tmp]))
62
+ df_stats_test = pd.DataFrame([stats_tmp])
63
+ print(df_stats_test.to_markdown(index=False))
64
+
t_rex_relation_similarity.py ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import json
2
+ from itertools import product
3
+
4
+ import datasets
5
+
6
+
7
+ logger = datasets.logging.get_logger(__name__)
8
+ _DESCRIPTION = """T-Rex dataset."""
9
+ _NAME = "t_rex_relation_similarity"
10
+ _VERSION = "0.0.0"
11
+ _CITATION = """
12
+ @inproceedings{elsahar2018t,
13
+ title={T-rex: A large scale alignment of natural language with knowledge base triples},
14
+ author={Elsahar, Hady and Vougiouklis, Pavlos and Remaci, Arslen and Gravier, Christophe and Hare, Jonathon and Laforest, Frederique and Simperl, Elena},
15
+ booktitle={Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
16
+ year={2018}
17
+ }
18
+ """
19
+
20
+ _HOME_PAGE = "https://github.com/asahi417/relbert"
21
+ _URL = f'https://huggingface.co/datasets/relbert/{_NAME}/resolve/main/data'
22
+ MIN_ENTITY_FREQ = [1, 2, 3, 4]
23
+ MAX_PREDICATE_FREQ = [100, 50, 25, 10]
24
+
25
+ _TYPES = [f"filter_unified.min_entity_{a}_max_predicate_{b}" for a, b in product(MIN_ENTITY_FREQ, MAX_PREDICATE_FREQ)]
26
+ _URLS = {i: {
27
+ str(datasets.Split.TRAIN): [f'{_URL}/t_rex.{i}.train.jsonl'],
28
+ str(datasets.Split.VALIDATION): [f'{_URL}/t_rex.{i}.validation.jsonl'],
29
+ str(datasets.Split.TEST): [f'{_URL}/t_rex.filter_unified.test.jsonl']
30
+ } for i in _TYPES}
31
+
32
+
33
+ class TREXRelationSimilarityConfig(datasets.BuilderConfig):
34
+ """BuilderConfig"""
35
+
36
+ def __init__(self, **kwargs):
37
+ """BuilderConfig.
38
+ Args:
39
+ **kwargs: keyword arguments forwarded to super.
40
+ """
41
+ super(TREXRelationSimilarityConfig, self).__init__(**kwargs)
42
+
43
+
44
+ class TREXRelationSimilarity(datasets.GeneratorBasedBuilder):
45
+ """Dataset."""
46
+
47
+ BUILDER_CONFIGS = [
48
+ TREXRelationSimilarityConfig(name=i, version=datasets.Version(_VERSION), description=_DESCRIPTION)
49
+ for i in sorted(_TYPES)
50
+ ]
51
+
52
+ def _split_generators(self, dl_manager):
53
+ downloaded_file = dl_manager.download_and_extract(_URLS[self.config.name])
54
+ return [datasets.SplitGenerator(name=i, gen_kwargs={"filepaths": downloaded_file[str(i)]})
55
+ for i in [datasets.Split.TRAIN, datasets.Split.VALIDATION, datasets.Split.TEST]]
56
+
57
+ def _generate_examples(self, filepaths):
58
+ _key = 0
59
+ for filepath in filepaths:
60
+ logger.info(f"generating examples from = {filepath}")
61
+ with open(filepath, encoding="utf-8") as f:
62
+ _list = [i for i in f.read().split('\n') if len(i) > 0]
63
+ for i in _list:
64
+ data = json.loads(i)
65
+ yield _key, data
66
+ _key += 1
67
+
68
+ def _info(self):
69
+ return datasets.DatasetInfo(
70
+ description=_DESCRIPTION,
71
+ features=datasets.Features(
72
+ {
73
+ "relation_type": datasets.Value("string"),
74
+ "positives": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
75
+ "negatives": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
76
+ }
77
+ ),
78
+ supervised_keys=None,
79
+ homepage=_HOME_PAGE,
80
+ citation=_CITATION,
81
+ )