drwiner / ScreenPy

Automated Screenplay Annotation for Extracting Storytelling Knowledge

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ScreenPy

Parse and annotate screenplays with markup language

File info
screenpile Main algorithms for compiling screenplay structure
screenpy Parses shot headings with recursive descent
screenpy_vars Defines valid shot heading element patterns
screenpy_tests Extra tests in addition to screenpy
moviescript_crawler crawls through imsdb_raw_nov_2015 folder
parsing_stats outputs table (below)

Dependencies:

PyParsing (http://pyparsing.wikispaces.com/) (for parsing shot heading grammar)

sense2vec (spaCy) (https://github.com/explosion/sense2vec) (for temporal expression classification)

Screenplay Structure

See the attached paper for more details (Paper)

Segmentation

IMSDb screenplay parsing

ScreenPy parses raw screenplays. The following table are the stats for a corpus of screenplays extracted from IMSDb (www.imsdb.com). Some screenplays are tagged as multiple genres.

Screenplay genre: number of films, avg number of master headings (msegs), avg number of segments (segs), avg subset of segments which have shot heading (headings), avg subset of segments which are speaker (speakers), avg number of headings which have a shot type (has shot), avg number of headings which have subject (has subj), and avg number of headings which have a time of day (has tod).

GENRE FILMS msegs segs headings speakers has shot has subj has tod
Action 272 145.32 1240.44 621.07 538.46 26.28 207.86 89.86
Adventure 143 135.96 1201.85 574.01 546.62 19.32 181.02 80.64
Animation 24 83.42 1190.79 466.08 674.71 18.96 116.88 47.92
Biography 3 144.33 1352.67 543.00 756.00 8.33 57.67 123.00
Comedy 310 114.37 1370.38 581.92 720.32 20.25 174.33 78.83
Crime 193 135.07 1352.31 656.12 620.73 18.33 185.67 93.49
Drama 541 123.10 1328.10 591.28 666.60 20.36 161.21 85.77
Family 22 110.45 1385.86 740.59 561.36 36.95 208.23 59.64
Fantasy 90 120.06 1213.94 627.40 526.83 29.60 227.74 73.33
Film-Noir 4 49.50 1535.00 721.75 786.50 139.25 201.50 13.00
History 3 107.33 1302.00 594.00 600.67 1.67 175.33 95.00
Horror 134 124.38 1149.94 631.57 450.96 32.89 238.72 79.32
Music 5 110.40 1179.20 503.40 660.80 7.00 53.60 96.40
Musical 14 96.21 1388.93 712.64 606.29 27.00 208.43 71.57
Mystery 97 137.91 1294.98 608.71 621.03 14.27 168.92 90.93
Romance 177 116.37 1367.10 596.67 710.13 25.08 176.86 84.27
Sci-Fi 140 142.34 1161.19 606.56 471.83 23.66 202.91 79.14
Short 3 39.00 269.33 138.33 118.00 8.33 32.67 25.67
Sport 2 229.00 1380.00 881.00 499.00 3.00 647.00 71.00
Thriller 352 135.31 1265.74 626.78 574.90 25.98 216.25 86.01
War 25 105.76 1272.40 616.08 577.88 26.36 263.68 85.88
Western 10 145.70 1367.10 706.00 602.10 76.10 223.50 103.30
*** *** *** *** *** *** *** *** ***
sum 2564 2651.29 27569.26 13344.96 12891.72 608.97 4329.98 1713.95
avg 116.55 120.51 1253.15 606.59 585.99 27.68 196.82 77.91

Verb sense disambiguation

genre num avg SD frames / screenplay avg frames / mseg avg frames / seg avg msegs / screenplay avg segs / screenplay avg headings/screenplay avg speaking-turns/ screenplay avg has-shot avg has-shot has-subj
Western 10 1442.10 9.90 1.05 145.7 1367.1 706 602.1 76.1 223.5 103.3
Film-Noir 4 1353.00 27.33 0.88 49.5 1535 721.75 786.5 139.25 201.5 13
Sport 2 1170.00 5.11 0.85 229 1380 881 499 3 647 71
Horror 134 1117.10 8.98 0.97 124.380597 1149.940299 631.5671642 450.9552239 32.8880597 238.7238806 79.32089552
Mystery 97 1092.80 7.92 0.84 137.9072165 1294.979381 608.7113402 621.0309278 14.26804124 168.9175258 90.92783505
Thriller 352 1078.94 7.97 0.85 135.3068182 1265.744318 626.7755682 574.9005682 25.98011364 216.2528409 86.00568182
Crime 193 1074.30 7.95 0.79 135.0673575 1352.310881 656.1243523 620.7253886 18.33160622 185.6683938 93.48704663
Sci-Fi 140 1062.73 7.47 0.92 142.3357143 1161.192857 606.5571429 471.8285714 23.65714286 202.9142857 79.13571429
War 25 1061.08 10.03 0.83 105.76 1272.4 616.08 577.88 26.36 263.68 85.88
Fantasy 90 1056.44 8.80 0.87 120.0555556 1213.944444 627.4 526.8333333 29.6 227.7444444 73.33333333
Action 272 1053.60 7.25 0.85 145.3235294 1240.441176 621.0735294 538.4632353 26.28308824 207.8566176 89.86397059
Family 22 1052.23 9.53 0.76 110.4545455 1385.863636 740.5909091 561.3636364 36.95454545 208.2272727 59.63636364
Music 5 1033.60 9.36 0.88 110.4 1179.2 503.4 660.8 7 53.6 96.4
Adventure 143 1008.74 7.42 0.84 135.958042 1201.846154 574.006993 546.6223776 19.32167832 181.020979 80.64335664
Musical 14 996.86 10.36 0.72 96.21428571 1388.928571 712.6428571 606.2857143 27 208.4285714 71.57142857
Drama 541 996.49 8.09 0.75 123.103512 1328.096118 591.2809612 666.6044362 20.35674677 161.2144177 85.77264325
Romance 177 969.79 8.33 0.71 116.3672316 1367.101695 596.6666667 710.1299435 25.08474576 176.8587571 84.26553672
Biography 3 965.33 6.69 0.71 144.3333333 1352.666667 543 756 8.333333333 57.66666667 123
Comedy 310 899.93 7.87 0.66 114.3741935 1370.380645 581.9193548 720.3225806 20.24516129 174.3290323 78.82580645
History 3 860.00 8.01 0.66 107.3333333 1302 594 600.6666667 1.666666667 175.3333333 95
Animation 24 768.71 9.22 0.65 83.41666667 1190.791667 466.0833333 674.7083333 18.95833333 116.875 47.91666667
Short 3 189.67 4.86 0.70 39 269.3333333 138.3333333 118 8.333333333 32.66666667 25.66666667

Development

Working on verb sense disambiguation to classify clauses in stage direction with FrameNet frames using off-the-shelf frame-semantic parser [Semafor] (https://github.com/Noahs-ARK/semafor).

About

Automated Screenplay Annotation for Extracting Storytelling Knowledge


Languages

Language:Python 100.0%