blockers
DeepBlocker
Bases: EmbeddingBlocker
Base class for DeepBlocker strategies.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
frame_encoder |
HintOrType[DeepBlockerFrameEncoder]
|
DeepBlockerFrameEncoder: DeepBlocker strategy. |
None
|
frame_encoder_kwargs |
OptionalKwargs
|
keyword arguments for initialisation of encoder |
None
|
embedding_block_builder_kwargs |
OptionalKwargs
|
keyword arguments for initalising blockbuilder. |
None
|
save |
bool
|
If true saves the embeddings before using blockbuilding. |
True
|
save_dir |
Optional[Union[str, Path]]
|
Directory where to save the embeddings. |
None
|
force |
bool
|
If true, recalculate the embeddings and overwrite existing. Else use precalculated if present. |
False
|
Attributes:
Name | Type | Description |
---|---|---|
frame_encoder |
DeepBlocker Encoder class to use for embedding the datasets. |
|
embedding_block_builder |
Block building class to create blocks from embeddings. |
|
save |
If true saves the embeddings before using blockbuilding. |
|
save_dir |
Directory where to save the embeddings. |
|
force |
If true, recalculate the embeddings and overwrite existing. Else use precalculated if present. |
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import DeepBlocker
>>> blocker = DeepBlocker(frame_encoder="autoencoder")
>>> blocks = blocker.assign(left=ds.left, right=ds.right)
Reference
Thirumuruganathan et. al. 'Deep Learning for Blocking in Entity Matching: A Design Space Exploration', VLDB 2021, http://vldb.org/pvldb/vol14/p2459-thirumuruganathan.pdf
Source code in klinker/blockers/embedding/deepblocker.py
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 |
|
EmbeddingBlocker
Bases: SchemaAgnosticBlocker
Base class for embedding-based blocking approaches.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
frame_encoder |
HintOrType[FrameEncoder]
|
Encoder class to use for embedding the datasets. |
None
|
frame_encoder_kwargs |
OptionalKwargs
|
keyword arguments for initialising encoder class. |
None
|
embedding_block_builder |
HintOrType[EmbeddingBlockBuilder]
|
Block building class to create blocks from embeddings. |
None
|
embedding_block_builder_kwargs |
OptionalKwargs
|
keyword arguments for initalising blockbuilder. |
None
|
save |
bool
|
If true saves the embeddings before using blockbuilding. |
True
|
save_dir |
Optional[Union[str, Path]]
|
Directory where to save the embeddings. |
None
|
force |
bool
|
If true, recalculate the embeddings and overwrite existing. Else use precalculated if present. |
False
|
Attributes:
Name | Type | Description |
---|---|---|
frame_encoder |
Encoder class to use for embedding the datasets. |
|
embedding_block_builder |
Block building class to create blocks from embeddings. |
|
save |
If true saves the embeddings before using blockbuilding. |
|
save_dir |
Directory where to save the embeddings. |
|
force |
If true, recalculate the embeddings and overwrite existing. Else use precalculated if present. |
Source code in klinker/blockers/embedding/blocker.py
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 |
|
from_encoded(left_path=None, right_path=None, left_name=None, right_name=None)
Apply blockbuilding strategy from precalculated embeddings.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
left_path |
path of left encoding. |
None
|
|
right_path |
path of right encoding. |
None
|
|
left_name |
Name of left dataset. |
None
|
|
right_name |
Name of right dataset. |
None
|
Returns:
Type | Description |
---|---|
KlinkerBlockManager
|
Calculated blocks. |
Source code in klinker/blockers/embedding/blocker.py
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 |
|
save_encoded(save_dir, encodings, table_names)
staticmethod
Save embeddings.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
save_dir |
Union[str, Path]
|
Union[str, pathlib.Path]: Directory to save into. |
required |
encodings |
Tuple[NamedVector, NamedVector]
|
Tuple[NamedVector, NamedVector]: Tuple of named embeddings. |
required |
table_names |
Tuple[str, str]
|
Tuple[str, str]: Name of left/right dataset. |
required |
Source code in klinker/blockers/embedding/blocker.py
144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 |
|
MinHashLSHBlocker
Bases: SchemaAgnosticBlocker
Blocker relying on MinHashLSH procedure.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
tokenize_fn |
Callable
|
Function that tokenizes entity attribute values. |
word_tokenize
|
threshold |
float
|
float: Jaccard threshold to use in underlying lsh procedure. |
0.5
|
num_perm |
int
|
int: number of permutations used in minhash algorithm. |
128
|
weights |
Tuple[float, float]
|
Tuple[float,float]: false positive/false negative weighting (must add up to one) |
(0.5, 0.5)
|
Attributes:
Name | Type | Description |
---|---|---|
tokenize_fn |
Callable
|
Function that tokenizes entity attribute values. |
threshold |
float: Jaccard threshold to use in underlying lsh procedure. |
|
num_perm |
int: number of permutations used in minhash algorithm. |
|
weights |
Tuple[float,float]: false positive/false negative weighting (must add up to one) |
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import MinHashLSHBlocker
>>> blocker = MinHashLSHBlocker(threshold=0.8, weights=(0.7,0.3))
>>> blocks = blocker.assign(left=ds.left, right=ds.right)
Source code in klinker/blockers/lsh.py
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 |
|
QgramsBlocker
Bases: StandardBlocker
Blocker relying on qgram procedure
Parameters:
Name | Type | Description | Default |
---|---|---|---|
blocking_key |
str
|
str: On which attribute the blocking should be done |
required |
q |
int
|
int: how big the qgrams should be. |
3
|
Attributes:
Name | Type | Description |
---|---|---|
blocking_key |
str: On which attribute the blocking should be done |
|
q |
int: how big the qgrams should be. |
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import QgramsBlocker
>>> blocker = QgramsBlocker(blocking_key="tail")
>>> blocks = blocker.assign(left=ds.left, right=ds.right)
Source code in klinker/blockers/qgrams.py
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 |
|
assign(left, right, left_rel=None, right_rel=None)
Assign entity ids to blocks.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
left |
KlinkerFrame
|
KlinkerFrame: Contains entity attribute information of left dataset. |
required |
right |
KlinkerFrame
|
KlinkerFrame: Contains entity attribute information of right dataset. |
required |
left_rel |
Optional[KlinkerFrame]
|
Optional[KlinkerFrame]: (Default value = None) Contains relational information of left dataset. |
None
|
right_rel |
Optional[KlinkerFrame]
|
Optional[KlinkerFrame]: (Default value = None) Contains relational information of left dataset. |
None
|
Returns:
Name | Type | Description |
---|---|---|
KlinkerBlockManager |
KlinkerBlockManager
|
instance holding the resulting blocks. |
Source code in klinker/blockers/qgrams.py
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 |
|
qgram_tokenize(x)
Tokenize into qgrams
Parameters:
Name | Type | Description | Default |
---|---|---|---|
x |
str
|
str: input string |
required |
Returns:
Type | Description |
---|---|
Optional[List[str]]
|
list of qgrams |
Source code in klinker/blockers/qgrams.py
36 37 38 39 40 41 42 43 44 45 46 47 48 |
|
RelationalDeepBlocker
Bases: RelationalBlocker
Seperate DeepBlocker strategy on concatenation of entity attribute values and neighboring values.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import RelationalDeepBlocker
>>> blocker = RelationalDeepBlocker(attr_frame_encoder="autoencoder", rel_frame_encoder="autoencoder")
>>> blocks = blocker.assign(left=ds.left, right=ds.right, left_rel=ds.left_rel, right_rel=ds.right_rel)
Source code in klinker/blockers/relation_aware.py
335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 |
|
RelationalMinHashLSHBlocker
Bases: RelationalBlocker
Seperate MinHashLSH blocking on concatenation of entity attribute values and neighboring values.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import RelationalMinHashLSHBlocker
>>> blocker = RelationalMinHashLSHBlocker(attr_threshold=0.7, rel_threshold=0.9)
>>> blocks = blocker.assign(left=ds.left, right=ds.right, left_rel=ds.left_rel, right_rel=ds.right_rel)
Source code in klinker/blockers/relation_aware.py
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 |
|
RelationalTokenBlocker
Bases: RelationalBlocker
Seperate Tokenblocking on concatenation of entity attribute values and neighboring values.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import RelationalTokenBlocker
>>> blocker = RelationalTokenBlocker(attr_min_token_length=3, rel_min_token_length=5)
>>> blocks = blocker.assign(left=ds.left, right=ds.right, left_rel=ds.left_rel, right_rel=ds.right_rel)
Source code in klinker/blockers/relation_aware.py
304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 |
|
SimpleRelationalMinHashLSHBlocker
Bases: BaseSimpleRelationalBlocker
MinHashLSH blocking on concatenation of entity attribute values and neighboring values.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import SimpleRelationalTokenBlocker
>>> blocker = SimpleRelationalMinHashLSHBlocker()
>>> blocks = blocker.assign(left=ds.left, right=ds.right, left_rel=ds.left_rel, right_rel=ds.right_rel)
Source code in klinker/blockers/relation_aware.py
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 |
|
SimpleRelationalTokenBlocker
Bases: BaseSimpleRelationalBlocker
Token blocking on concatenation of entity attribute values and neighboring values.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import SimpleRelationalTokenBlocker
>>> blocker = SimpleRelationalTokenBlocker()
>>> blocks = blocker.assign(left=ds.left, right=ds.right, left_rel=ds.left_rel, right_rel=ds.right_rel)
Source code in klinker/blockers/relation_aware.py
172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 |
|
StandardBlocker
Bases: Blocker
Block on same values of a specific column.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import StandardBlocker
>>> blocker = StandardBlocker(blocking_key="tail")
>>> blocks = blocker.assign(left=ds.left, right=ds.right)
Reference
Fellegi, Ivan P. and Alan B. Sunter. 'A Theory for Record Linkage.' Journal of the American Statistical Association 64 (1969): 1183-1210.
Source code in klinker/blockers/standard.py
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 |
|
assign(left, right, left_rel=None, right_rel=None)
Assign entity ids to blocks.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
left |
KlinkerFrame
|
KlinkerFrame: Contains entity attribute information of left dataset. |
required |
right |
KlinkerFrame
|
KlinkerFrame: Contains entity attribute information of right dataset. |
required |
left_rel |
Optional[KlinkerFrame]
|
Optional[KlinkerFrame]: (Default value = None) Contains relational information of left dataset. |
None
|
right_rel |
Optional[KlinkerFrame]
|
Optional[KlinkerFrame]: (Default value = None) Contains relational information of left dataset. |
None
|
Returns:
Name | Type | Description |
---|---|---|
KlinkerBlockManager |
KlinkerBlockManager
|
instance holding the resulting blocks. |
Source code in klinker/blockers/standard.py
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 |
|
TokenBlocker
Bases: SchemaAgnosticBlocker
Concatenates and tokenizes entity attribute values and blocks on tokens.
Examples:
>>> # doctest: +SKIP
>>> from sylloge import MovieGraphBenchmark
>>> from klinker.data import KlinkerDataset
>>> ds = KlinkerDataset.from_sylloge(MovieGraphBenchmark(),clean=True)
>>> from klinker.blockers import TokenBlocker
>>> blocker = TokenBlocker()
>>> blocks = blocker.assign(left=ds.left, right=ds.right)
Source code in klinker/blockers/token_blocking.py
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 |
|