ikarasz commited on
Commit
3d37597
·
1 Parent(s): 0b9b444

add math words

Browse files
Files changed (2) hide show
  1. handler.py +2 -2
  2. utils.py +421 -0
handler.py CHANGED
@@ -378,8 +378,8 @@ def load_math_terms():
378
  math_terms_dict = {}
379
  for term in MATH_WORDS:
380
  if term in MATH_PREFIXES:
381
- math_terms_dict[f"(^|[^a-zA-Z]){term}(s|es)?([^a-zA-Z]|$)"] = term
382
- math_terms.append(f"(^|[^a-zA-Z]){term}(s|es)?([^a-zA-Z]|$)")
383
  else:
384
  math_terms.append(term)
385
  math_terms_dict[term] = term
 
378
  math_terms_dict = {}
379
  for term in MATH_WORDS:
380
  if term in MATH_PREFIXES:
381
+ math_terms_dict[f"(^|[^a-zA-Z]){term}(s|es|d|ed)?([^a-zA-Z]|$)"] = term
382
+ math_terms.append(f"(^|[^a-zA-Z]){term}(s|es|d|ed)?([^a-zA-Z]|$)")
383
  else:
384
  math_terms.append(term)
385
  math_terms_dict[term] = term
utils.py CHANGED
@@ -295,6 +295,427 @@ MATH_WORDS = [
295
  "unit rate",
296
  "variability",
297
  "variable",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
298
  ]
299
 
300
  def get_num_words(text):
 
295
  "unit rate",
296
  "variability",
297
  "variable",
298
+ "abundant number",
299
+ "accurate",
300
+ "acre",
301
+ "addition fact",
302
+ "algebraic",
303
+ "altitude",
304
+ "apex",
305
+ "arithmetic facts",
306
+ "associative property",
307
+ "astronomical unit",
308
+ "base",
309
+ "baseline",
310
+ "billion",
311
+ "celsius",
312
+ "census",
313
+ "cent",
314
+ "center of a circle",
315
+ "center of a sphere",
316
+ "chance",
317
+ "circle graph",
318
+ "column",
319
+ "combine",
320
+ "common fraction",
321
+ "comparison diagram",
322
+ "comparison story",
323
+ "compass",
324
+ "complement",
325
+ "concave polygon",
326
+ "concentric circles",
327
+ "consecutive",
328
+ "constant",
329
+ "continuous model of area",
330
+ "continuous model of volume",
331
+ "contour",
332
+ "conversion fact",
333
+ "convex polygon",
334
+ "counting numbers",
335
+ "counting up subtraction",
336
+ "cover-up method",
337
+ "cross multiplication",
338
+ "cubic",
339
+ "cubit",
340
+ "curved surface",
341
+ "cylinder",
342
+ "decagon",
343
+ "decimeter",
344
+ "deficient number",
345
+ "density",
346
+ "discrete model",
347
+ "displacement method",
348
+ "divisibility test",
349
+ "divisible by",
350
+ "dodecahedron",
351
+ "double stem plot",
352
+ "doubles fact",
353
+ "egyptian multiplication",
354
+ "elevation",
355
+ "embed figure",
356
+ "end point",
357
+ "enlarge",
358
+ "equal",
359
+ "equal groups",
360
+ "equal parts",
361
+ "equidistant marks",
362
+ "equilateral polygon",
363
+ "equivalent fractions",
364
+ "european subtraction",
365
+ "expanded notation",
366
+ "expected outcome",
367
+ "exponential",
368
+ "extended facts",
369
+ "fact power",
370
+ "fact triangle",
371
+ "factor",
372
+ "factors of numbers",
373
+ "fahrenheit",
374
+ "false number sentence",
375
+ "figurate numbers",
376
+ "flowchart",
377
+ "fluid ounce",
378
+ "fractional part",
379
+ "fulcrum",
380
+ "function machine",
381
+ "furlong",
382
+ "genus",
383
+ "geoboard",
384
+ "geometric solid",
385
+ "geometry template",
386
+ "girth",
387
+ "golden ratio",
388
+ "golden rectangle",
389
+ "graph key",
390
+ "grouping symbol",
391
+ "hemisphere",
392
+ "icosahedron",
393
+ "improper fraction",
394
+ "inch",
395
+ "index of locations",
396
+ "indirect measurement",
397
+ "input",
398
+ "inscribed polygon",
399
+ "instance of a pattern",
400
+ "interior of a figure",
401
+ "interpolate",
402
+ "irrational",
403
+ "isometry transformation",
404
+ "isosceles trapezoid",
405
+ "juxtapose",
406
+ "key sequence",
407
+ "label",
408
+ "landmark",
409
+ "latitude",
410
+ "lattice multiplication",
411
+ "left to right subtraction",
412
+ "leg of a right triangle",
413
+ "like terms",
414
+ "line graph",
415
+ "line of reflection",
416
+ "line of symmetry",
417
+ "line symmetry",
418
+ "lines of latitude",
419
+ "lines of longitude",
420
+ "longitude",
421
+ "magnitude estimate",
422
+ "map legend",
423
+ "map scale",
424
+ "maximum",
425
+ "measurement division",
426
+ "measurement unit",
427
+ "meridian bar",
428
+ "metric system",
429
+ "midpoint",
430
+ "mile",
431
+ "millisecond",
432
+ "minimum",
433
+ "minuend",
434
+ "mirror image",
435
+ "mobius",
436
+ "modal",
437
+ "multiplication counting principle",
438
+ "multiplication diagram",
439
+ "multiplication fact",
440
+ "multiplication symbols",
441
+ "multiplication use class",
442
+ "negative rational numbers",
443
+ "nested parentheses",
444
+ "net score",
445
+ "net weight",
446
+ "nonagon",
447
+ "nonconvex polygon",
448
+ "normal span",
449
+ "number grid",
450
+ "number sequence",
451
+ "numeral",
452
+ "numeration",
453
+ "octahedron",
454
+ "open proportion",
455
+ "operation",
456
+ "operation symbol",
457
+ "opposite angle",
458
+ "opposite change rule",
459
+ "opposite of a number",
460
+ "opposite side",
461
+ "order of magnitude",
462
+ "order of operations",
463
+ "order of rotation symmetry",
464
+ "ordinal number",
465
+ "pan balance",
466
+ "parabola",
467
+ "parallel lines",
468
+ "parallel planes",
469
+ "part to part ratio",
470
+ "part to whole ratio",
471
+ "part whole fraction",
472
+ "partial differences subtraction",
473
+ "partial products multiplication",
474
+ "partial quotients division",
475
+ "partial sums addition",
476
+ "partitive division",
477
+ "parts and total diagram",
478
+ "per capita",
479
+ "per unit rate",
480
+ "percent circle",
481
+ "perfect number",
482
+ "perpetual calendar",
483
+ "pie graph",
484
+ "plane",
485
+ "plane figure",
486
+ "point symmetry",
487
+ "population density",
488
+ "precise",
489
+ "predict",
490
+ "prediction line",
491
+ "preimage",
492
+ "prime factorization",
493
+ "prime meridian",
494
+ "probability",
495
+ "probability meter",
496
+ "probability tree diagram",
497
+ "proper factor",
498
+ "proper fraction",
499
+ "property",
500
+ "quadrangle",
501
+ "quick common denominator",
502
+ "quotitive division",
503
+ "random draw",
504
+ "random experiment",
505
+ "random number",
506
+ "random sample",
507
+ "rank",
508
+ "rate diagram",
509
+ "rate multiplication ",
510
+ "rate unit",
511
+ "recall survey",
512
+ "rectangular array",
513
+ "rectangular coordinate grid",
514
+ "rectangular prism",
515
+ "rectangular pyramid",
516
+ "rectilinear figure",
517
+ "reflection",
518
+ "reflex angle",
519
+ "regular polyhedron",
520
+ "regular tessellation",
521
+ "relation symbol",
522
+ "revolution",
523
+ "right cone",
524
+ "right cylinder",
525
+ "right prism",
526
+ "right pyramid",
527
+ "right triangle",
528
+ "roman numerals",
529
+ "rotation symmetry",
530
+ "same change rule for subtraction",
531
+ "scale model",
532
+ "scale of a map",
533
+ "scale of a number line",
534
+ "sector",
535
+ "segment",
536
+ "sequence",
537
+ "significant digits",
538
+ "similar figures",
539
+ "simpler form",
540
+ "situtation diagram",
541
+ "skew lines",
542
+ "slanted",
543
+ "slide rule",
544
+ "span",
545
+ "stacked bar graph",
546
+ "standard unit",
547
+ "stem and leaf plot",
548
+ "step graph",
549
+ "straightedge",
550
+ "substitute",
551
+ "subtrahend",
552
+ "surface",
553
+ "symmetric",
554
+ "tally",
555
+ "tangent",
556
+ "tangent circles",
557
+ "temperature",
558
+ "template",
559
+ "tetrahedron",
560
+ "theorem",
561
+ "tile",
562
+ "tiling",
563
+ "time graph",
564
+ "timeline",
565
+ "top heavy fraction",
566
+ "topological",
567
+ "topology",
568
+ "trade first subtraction",
569
+ "tree diagram",
570
+ "triangular",
571
+ "true number sentence",
572
+ "truncate",
573
+ "twin primes",
574
+ "unlike denominators",
575
+ "unlike fractions",
576
+ "vanishing ",
577
+ "venn diagram",
578
+ "vernal equinox",
579
+ "weight",
580
+ "width",
581
+ "base of a prism",
582
+ "base of a pyramid",
583
+ "face",
584
+ "numerical data",
585
+ "opposite",
586
+ "pace",
587
+ "per",
588
+ "region",
589
+ "sign",
590
+ "alternate interior angles",
591
+ "base of an exponent",
592
+ "cone",
593
+ "congruent",
594
+ "counterclockwise",
595
+ "cube root",
596
+ "hypotenuse",
597
+ "irrational number",
598
+ "linear relationship",
599
+ "positive association",
600
+ "rate of change",
601
+ "translation",
602
+ "transversal",
603
+ "circumference",
604
+ "corresponding",
605
+ "expand",
606
+ "population",
607
+ "proportion",
608
+ "radius",
609
+ "random",
610
+ "repeating decimal",
611
+ "representative",
612
+ "scaled",
613
+ "withdrawal",
614
+ "center",
615
+ "edge",
616
+ "height of a parallelogram or triangle",
617
+ "net",
618
+ "speed",
619
+ "table",
620
+ "term",
621
+ "adjacent",
622
+ "complementary",
623
+ "cross-section",
624
+ "cross section",
625
+ "deposit",
626
+ "event",
627
+ "measurement error",
628
+ "proportional",
629
+ "simulation",
630
+ "center of a dilation",
631
+ "clockwise",
632
+ "dilation",
633
+ "function",
634
+ "negative association",
635
+ "pythagorean theorem",
636
+ "relative frequency",
637
+ "rigid transformation",
638
+ "scale factor",
639
+ "scatter plot",
640
+ "similar",
641
+ "sphere",
642
+ "two-way table",
643
+ "additive identity",
644
+ "additive inverse",
645
+ "box and whisker plot",
646
+ "cartesian coordinates",
647
+ "central angle",
648
+ "chord",
649
+ "combination",
650
+ "commutative property",
651
+ "coplanar",
652
+ "cross product",
653
+ "dependent events",
654
+ "difference",
655
+ "dividend",
656
+ "equilateral triangle",
657
+ "error of measurement",
658
+ "factorial",
659
+ "formula",
660
+ "identity property of",
661
+ "independent events",
662
+ "infinity",
663
+ "inscribed angle",
664
+ "intercept",
665
+ "intercepted arc",
666
+ "inverse",
667
+ "inverse operations",
668
+ "isosceles triangle",
669
+ "least common denominator",
670
+ "like fractions",
671
+ "locus",
672
+ "logic",
673
+ "lowest terms",
674
+ "mode",
675
+ "multiplicative identity",
676
+ "multiplicative inverse",
677
+ "mutually exclusive events",
678
+ "natural numbers",
679
+ "normal",
680
+ "permutation",
681
+ "pi",
682
+ "point",
683
+ "power",
684
+ "range",
685
+ "rate",
686
+ "ray",
687
+ "real numbers",
688
+ "rectangular",
689
+ "root",
690
+ "rotation",
691
+ "scalene triangle",
692
+ "scattergram",
693
+ "set",
694
+ "statistics",
695
+ "terminating decimal",
696
+ "transformation",
697
+ "x intercept",
698
+ "x-axis",
699
+ "x-intercept",
700
+ "y intercept",
701
+ "y-axis",
702
+ "y-intercept",
703
+ "zero",
704
+ "zero property of multiplication",
705
+ "base of a parallelogram",
706
+ "base of a triangle",
707
+ "height",
708
+ "chance experiment",
709
+ "diameter",
710
+ "mean",
711
+ "percentage",
712
+ "sample",
713
+ "legs",
714
+ "outlier",
715
+ "slope",
716
+ "square root",
717
+ "system of equations",
718
+ "tessellation",
719
  ]
720
 
721
  def get_num_words(text):