vedant2905/CrossLingual
0
1{2 "16": {3 "aligned_clusters": [4 4195 ],6 "metrics": {7 "match_percentage": 0.6666666666666666,8 "source_cluster_size": 3,9 "target_cluster_sizes": [10 711 ],12 "aligned_word_count": 2,13 "total_words": 3,14 "size_threshold": 0.4,15 "translation_threshold": 0.5,16 "calign_score": 0.6666666666666666,17 "colap_score": 0.95584016879268518 }19 },20 "49": {21 "aligned_clusters": [22 136,23 11,24 175,25 476,26 15727 ],28 "metrics": {29 "match_percentage": 0.6666666666666666,30 "source_cluster_size": 6,31 "target_cluster_sizes": [32 9,33 7,34 8,35 12,36 737 ],38 "aligned_word_count": 4,39 "total_words": 6,40 "size_threshold": 0.4,41 "translation_threshold": 0.5,42 "calign_score": 0.8333333333333334,43 "colap_score": 0.934852784557121344 }45 },46 "92": {47 "aligned_clusters": [48 384,49 485,50 46251 ],52 "metrics": {53 "match_percentage": 0.6666666666666666,54 "source_cluster_size": 3,55 "target_cluster_sizes": [56 3,57 4,58 759 ],60 "aligned_word_count": 2,61 "total_words": 3,62 "size_threshold": 0.4,63 "translation_threshold": 0.5,64 "calign_score": 0.6666666666666666,65 "colap_score": 1.066 }67 },68 "104": {69 "aligned_clusters": [70 2,71 136,72 138,73 11,74 15,75 28,76 157,77 159,78 289,79 290,80 419,81 293,82 295,83 427,84 313,85 64,86 322,87 67,88 69,89 476,90 226,91 356,92 359,93 109,94 367,95 120,96 12297 ],98 "metrics": {99 "match_percentage": 0.75,100 "source_cluster_size": 4,101 "target_cluster_sizes": [102 5,103 9,104 9,105 4,106 8,107 6,108 8,109 9,110 2,111 3,112 7,113 8,114 8,115 6,116 6,117 8,118 5,119 7,120 9,121 8,122 6,123 6,124 4,125 5,126 7,127 3,128 7129 ],130 "aligned_word_count": 3,131 "total_words": 4,132 "size_threshold": 0.4,133 "translation_threshold": 0.5,134 "calign_score": 0.75,135 "colap_score": 0.9283310163047603136 }137 },138 "140": {139 "aligned_clusters": [140 280,141 476,142 30143 ],144 "metrics": {145 "match_percentage": 0.7142857142857143,146 "source_cluster_size": 7,147 "target_cluster_sizes": [148 16,149 17,150 7151 ],152 "aligned_word_count": 5,153 "total_words": 7,154 "size_threshold": 0.4,155 "translation_threshold": 0.5,156 "calign_score": 0.7142857142857143,157 "colap_score": 0.9325274083480994158 }159 },160 "154": {161 "aligned_clusters": [162 419,163 294,164 136,165 430,166 473,167 476168 ],169 "metrics": {170 "match_percentage": 0.6666666666666666,171 "source_cluster_size": 3,172 "target_cluster_sizes": [173 7,174 6,175 7,176 5,177 4,178 7179 ],180 "aligned_word_count": 2,181 "total_words": 3,182 "size_threshold": 0.4,183 "translation_threshold": 0.5,184 "calign_score": 0.6666666666666666,185 "colap_score": 0.9354026393899699186 }187 },188 "159": {189 "aligned_clusters": [190 382191 ],192 "metrics": {193 "match_percentage": 1.0,194 "source_cluster_size": 2,195 "target_cluster_sizes": [196 4197 ],198 "aligned_word_count": 2,199 "total_words": 2,200 "size_threshold": 0.4,201 "translation_threshold": 0.5,202 "calign_score": 1.0,203 "colap_score": 1.0204 }205 },206 "184": {207 "aligned_clusters": [208 293,209 37,210 487,211 153,212 383213 ],214 "metrics": {215 "match_percentage": 0.6666666666666666,216 "source_cluster_size": 3,217 "target_cluster_sizes": [218 5,219 6,220 7,221 7,222 4223 ],224 "aligned_word_count": 2,225 "total_words": 3,226 "size_threshold": 0.4,227 "translation_threshold": 0.5,228 "calign_score": 0.6666666666666666,229 "colap_score": 0.9999999999999998230 }231 },232 "189": {233 "aligned_clusters": [234 257,235 2,236 67,237 196,238 484,239 265,240 56241 ],242 "metrics": {243 "match_percentage": 0.6666666666666666,244 "source_cluster_size": 3,245 "target_cluster_sizes": [246 5,247 6,248 6,249 5,250 5,251 5,252 5253 ],254 "aligned_word_count": 2,255 "total_words": 3,256 "size_threshold": 0.4,257 "translation_threshold": 0.5,258 "calign_score": 0.6666666666666666,259 "colap_score": 1.0260 }261 },262 "193": {263 "aligned_clusters": [264 384265 ],266 "metrics": {267 "match_percentage": 0.6666666666666666,268 "source_cluster_size": 3,269 "target_cluster_sizes": [270 3271 ],272 "aligned_word_count": 2,273 "total_words": 3,274 "size_threshold": 0.4,275 "translation_threshold": 0.5,276 "calign_score": 0.6666666666666666,277 "colap_score": 0.9661712252069393278 }279 },280 "195": {281 "aligned_clusters": [282 2,283 67,284 226,285 290,286 293,287 322,288 136,289 356,290 359,291 419,292 364,293 427,294 367,295 120,296 122,297 476,298 28,299 159300 ],301 "metrics": {302 "match_percentage": 0.6666666666666666,303 "source_cluster_size": 3,304 "target_cluster_sizes": [305 5,306 4,307 6,308 2,309 3,310 7,311 6,312 6,313 5,314 7,315 6,316 6,317 4,318 6,319 5,320 7,321 3,322 7323 ],324 "aligned_word_count": 2,325 "total_words": 3,326 "size_threshold": 0.4,327 "translation_threshold": 0.5,328 "calign_score": 0.6666666666666666,329 "colap_score": 0.9999999999999998330 }331 },332 "196": {333 "aligned_clusters": [334 462335 ],336 "metrics": {337 "match_percentage": 1.0,338 "source_cluster_size": 2,339 "target_cluster_sizes": [340 4341 ],342 "aligned_word_count": 2,343 "total_words": 2,344 "size_threshold": 0.4,345 "translation_threshold": 0.5,346 "calign_score": 1.0,347 "colap_score": 1.0348 }349 },350 "216": {351 "aligned_clusters": [352 2,353 67,354 290,355 294,356 262,357 464358 ],359 "metrics": {360 "match_percentage": 0.6666666666666666,361 "source_cluster_size": 3,362 "target_cluster_sizes": [363 5,364 6,365 7,366 5,367 6,368 4369 ],370 "aligned_word_count": 2,371 "total_words": 3,372 "size_threshold": 0.4,373 "translation_threshold": 0.5,374 "calign_score": 0.6666666666666666,375 "colap_score": 0.9614241367387362376 }377 },378 "238": {379 "aligned_clusters": [380 322,381 356,382 359,383 42,384 363,385 177,386 409,387 91,388 476389 ],390 "metrics": {391 "match_percentage": 0.6666666666666666,392 "source_cluster_size": 3,393 "target_cluster_sizes": [394 4,395 5,396 5,397 6,398 6,399 4,400 7,401 4,402 7403 ],404 "aligned_word_count": 2,405 "total_words": 3,406 "size_threshold": 0.4,407 "translation_threshold": 0.5,408 "calign_score": 0.6666666666666666,409 "colap_score": 1.0410 }411 },412 "254": {413 "aligned_clusters": [414 293,415 37,416 487,417 153,418 383419 ],420 "metrics": {421 "match_percentage": 0.6666666666666666,422 "source_cluster_size": 3,423 "target_cluster_sizes": [424 5,425 6,426 7,427 7,428 4429 ],430 "aligned_word_count": 2,431 "total_words": 3,432 "size_threshold": 0.4,433 "translation_threshold": 0.5,434 "calign_score": 0.6666666666666666,435 "colap_score": 0.9999999999999998436 }437 },438 "265": {439 "aligned_clusters": [440 476,441 430,442 342,443 494444 ],445 "metrics": {446 "match_percentage": 0.6666666666666666,447 "source_cluster_size": 3,448 "target_cluster_sizes": [449 6,450 5,451 7,452 5453 ],454 "aligned_word_count": 2,455 "total_words": 3,456 "size_threshold": 0.4,457 "translation_threshold": 0.5,458 "calign_score": 0.6666666666666666,459 "colap_score": 0.9174685942852827460 }461 },462 "291": {463 "aligned_clusters": [464 297,465 45,466 115,467 54,468 348469 ],470 "metrics": {471 "match_percentage": 0.6666666666666666,472 "source_cluster_size": 3,473 "target_cluster_sizes": [474 7,475 3,476 5,477 5,478 7479 ],480 "aligned_word_count": 2,481 "total_words": 3,482 "size_threshold": 0.4,483 "translation_threshold": 0.5,484 "calign_score": 0.6666666666666666,485 "colap_score": 0.9169855238488014486 }487 },488 "305": {489 "aligned_clusters": [490 409,491 42,492 359493 ],494 "metrics": {495 "match_percentage": 1.0,496 "source_cluster_size": 2,497 "target_cluster_sizes": [498 4,499 4,500 4501 ],502 "aligned_word_count": 2,503 "total_words": 2,504 "size_threshold": 0.4,505 "translation_threshold": 0.5,506 "calign_score": 1.0,507 "colap_score": 1.0508 }509 },510 "336": {511 "aligned_clusters": [512 257,513 2,514 67,515 196,516 484,517 265,518 56519 ],520 "metrics": {521 "match_percentage": 0.6666666666666666,522 "source_cluster_size": 3,523 "target_cluster_sizes": [524 5,525 6,526 6,527 5,528 5,529 5,530 5531 ],532 "aligned_word_count": 2,533 "total_words": 3,534 "size_threshold": 0.4,535 "translation_threshold": 0.5,536 "calign_score": 0.6666666666666666,537 "colap_score": 0.961110569005777538 }539 },540 "350": {541 "aligned_clusters": [542 128,543 322,544 356,545 231,546 359,547 268,548 364,549 430,550 461,551 146,552 21,553 438,554 476,555 220,556 62557 ],558 "metrics": {559 "match_percentage": 0.6666666666666666,560 "source_cluster_size": 3,561 "target_cluster_sizes": [562 2,563 6,564 6,565 6,566 6,567 2,568 5,569 6,570 6,571 4,572 6,573 5,574 4,575 4,576 7577 ],578 "aligned_word_count": 2,579 "total_words": 3,580 "size_threshold": 0.4,581 "translation_threshold": 0.5,582 "calign_score": 0.6666666666666666,583 "colap_score": 0.974488826885052584 }585 },586 "364": {587 "aligned_clusters": [588 219,589 252,590 151591 ],592 "metrics": {593 "match_percentage": 0.6666666666666666,594 "source_cluster_size": 3,595 "target_cluster_sizes": [596 3,597 6,598 5599 ],600 "aligned_word_count": 2,601 "total_words": 3,602 "size_threshold": 0.4,603 "translation_threshold": 0.5,604 "calign_score": 0.6666666666666666,605 "colap_score": 0.9690253847702351606 }607 },608 "449": {609 "aligned_clusters": [610 191611 ],612 "metrics": {613 "match_percentage": 0.6666666666666666,614 "source_cluster_size": 3,615 "target_cluster_sizes": [616 7617 ],618 "aligned_word_count": 2,619 "total_words": 3,620 "size_threshold": 0.4,621 "translation_threshold": 0.5,622 "calign_score": 1.0,623 "colap_score": 0.9226816217067478624 }625 },626 "453": {627 "aligned_clusters": [628 226,629 419,630 191631 ],632 "metrics": {633 "match_percentage": 0.6666666666666666,634 "source_cluster_size": 3,635 "target_cluster_sizes": [636 7,637 6,638 7639 ],640 "aligned_word_count": 2,641 "total_words": 3,642 "size_threshold": 0.4,643 "translation_threshold": 0.5,644 "calign_score": 0.6666666666666666,645 "colap_score": 0.8972113169296332646 }647 },648 "457": {649 "aligned_clusters": [650 128,651 161,652 132,653 101,654 233,655 297,656 363,657 270,658 271,659 336,660 442,661 146,662 54,663 150,664 344,665 345,666 406,667 62668 ],669 "metrics": {670 "match_percentage": 0.6666666666666666,671 "source_cluster_size": 3,672 "target_cluster_sizes": [673 3,674 6,675 5,676 6,677 6,678 6,679 6,680 6,681 4,682 7,683 7,684 5,685 6,686 5,687 3,688 7,689 3,690 4691 ],692 "aligned_word_count": 2,693 "total_words": 3,694 "size_threshold": 0.4,695 "translation_threshold": 0.5,696 "calign_score": 0.6666666666666666,697 "colap_score": 1.0698 }699 },700 "458": {701 "aligned_clusters": [702 345703 ],704 "metrics": {705 "match_percentage": 1.0,706 "source_cluster_size": 2,707 "target_cluster_sizes": [708 3709 ],710 "aligned_word_count": 2,711 "total_words": 2,712 "size_threshold": 0.4,713 "translation_threshold": 0.5,714 "calign_score": 1.0,715 "colap_score": 1.0716 }717 },718 "473": {719 "aligned_clusters": [720 136,721 11,722 175,723 476,724 157725 ],726 "metrics": {727 "match_percentage": 0.8,728 "source_cluster_size": 5,729 "target_cluster_sizes": [730 9,731 7,732 8,733 12,734 7735 ],736 "aligned_word_count": 4,737 "total_words": 5,738 "size_threshold": 0.4,739 "translation_threshold": 0.5,740 "calign_score": 0.8,741 "colap_score": 0.9588597868656995742 }743 },744 "489": {745 "aligned_clusters": [746 367747 ],748 "metrics": {749 "match_percentage": 0.6666666666666666,750 "source_cluster_size": 3,751 "target_cluster_sizes": [752 5753 ],754 "aligned_word_count": 2,755 "total_words": 3,756 "size_threshold": 0.4,757 "translation_threshold": 0.5,758 "calign_score": 0.6666666666666666,759 "colap_score": 0.9222627814924376760 }761 }762}