diff --git a/.gitattributes b/.gitattributes index 0ae7c8e8d4df5634a14b5f652dccf1939edcefc2..b94a69be807b693e83f3049f28f579ef9b73330c 100644 --- a/.gitattributes +++ b/.gitattributes @@ -88,3 +88,21 @@ output/figures/hinglish_to_hindi_to_english/hinglish_to_hindi_to_english_\[hindi output/figures/hinglish_to_hindi_to_english/hinglish_to_hindi_to_english_\[hindi\]_roc.png filter=lfs diff=lfs merge=lfs -text output/figures/hinglish_to_hindi_to_english/hinglish_to_hindi_to_english_\[english\]_curves.png filter=lfs diff=lfs merge=lfs -text output/figures/hinglish_to_hindi_to_english/hinglish_to_hindi_to_english_\[Full\]_curves.png filter=lfs diff=lfs merge=lfs -text +shap/english_to_hindi_to_hinglish/shap_topwords_english.png filter=lfs diff=lfs merge=lfs -text +shap/english_to_hindi_to_hinglish/shap_topwords_full.png filter=lfs diff=lfs merge=lfs -text +shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.png filter=lfs diff=lfs merge=lfs -text +shap/english_to_hinglish_to_hindi/shap_topwords_english.png filter=lfs diff=lfs merge=lfs -text +shap/english_to_hinglish_to_hindi/shap_topwords_full.png filter=lfs diff=lfs merge=lfs -text +shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.png filter=lfs diff=lfs merge=lfs -text +shap/hindi_to_english_to_hinglish/shap_topwords_english.png filter=lfs diff=lfs merge=lfs -text +shap/hindi_to_english_to_hinglish/shap_topwords_full.png filter=lfs diff=lfs merge=lfs -text +shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.png filter=lfs diff=lfs merge=lfs -text +shap/hindi_to_hinglish_to_english/shap_topwords_english.png filter=lfs diff=lfs merge=lfs -text +shap/hindi_to_hinglish_to_english/shap_topwords_full.png filter=lfs diff=lfs merge=lfs -text +shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.png filter=lfs diff=lfs merge=lfs -text +shap/hinglish_to_english_to_hindi/shap_topwords_english.png filter=lfs diff=lfs merge=lfs -text +shap/hinglish_to_english_to_hindi/shap_topwords_full.png filter=lfs diff=lfs merge=lfs -text +shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.png filter=lfs diff=lfs merge=lfs -text +shap/hinglish_to_hindi_to_english/shap_topwords_english.png filter=lfs diff=lfs merge=lfs -text +shap/hinglish_to_hindi_to_english/shap_topwords_full.png filter=lfs diff=lfs merge=lfs -text +shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.png filter=lfs diff=lfs merge=lfs -text diff --git a/shap/english_to_hindi_to_hinglish/shap_summary.csv b/shap/english_to_hindi_to_hinglish/shap_summary.csv new file mode 100644 index 0000000000000000000000000000000000000000..710f433b2ccd641452b4c3357324c68e0b37fbb8 --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_summary.csv @@ -0,0 +1,5 @@ +model,eval_lang,top_hate_words,top_nonhate_words +english_to_hindi_to_hinglish,english,"blame(0.0227), cretin(0.0199), blaming(0.0151), unhelpful(0.0130), upwards(0.0121)","nt(-0.0291), wwf(-0.0279), facebook(-0.0261), ahh(-0.0256), cum(-0.0224)" +english_to_hindi_to_hinglish,hindi,"रे(0.0024), पढ़ने(0.0022), लाइव(0.0019), गाला(0.0017), औखत(0.0015)","आखिरकार(-0.0033), मुख(-0.0028), बिकुल(-0.0025), शायद(-0.0022), ह(-0.0021)" +english_to_hindi_to_hinglish,hinglish,"nawaz(0.0568), dhawan(0.0339), bashing(0.0253), shareef(0.0232), scn(0.0224)","gau(-0.0410), age(-0.0367), rajya(-0.0346), chori(-0.0336), channels(-0.0303)" +english_to_hindi_to_hinglish,full,"molvi(0.0300), chalo(0.0242), molana(0.0234), scn(0.0195), elitist(0.0188)","rajya(-0.0690), coding(-0.0279), meat(-0.0224), haan(-0.0219), maine(-0.0202)" diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_english.csv b/shap/english_to_hindi_to_hinglish/shap_topwords_english.csv new file mode 100644 index 0000000000000000000000000000000000000000..83dbfa0803f15454de5afa04374b02099ea2c1d2 --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_english.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +nt,-0.029088862240314484,non-hate,0.029088862240314484 +wwf,-0.02789453975856304,non-hate,0.02789453975856304 +facebook,-0.026072523556649685,non-hate,0.026072523556649685 +ahh,-0.025606895331293344,non-hate,0.025606895331293344 +blame,0.022735290229320526,hate,0.022735290229320526 +cum,-0.022419555112719536,non-hate,0.022419555112719536 +degenerate,-0.02214253880083561,non-hate,0.02214253880083561 +cretin,0.0198978204280138,hate,0.0198978204280138 +grow,-0.01629100926220417,non-hate,0.01629100926220417 +unamused,-0.01572409241149823,non-hate,0.01572409241149823 +chat,-0.015594650292769074,non-hate,0.015594650292769074 +passes,-0.015481927432119846,non-hate,0.015481927432119846 +perverted,-0.015152966603636742,non-hate,0.015152966603636742 +blaming,0.015061663463711739,hate,0.015061663463711739 +andy,-0.014559097588062286,non-hate,0.014559097588062286 +headings,-0.014278455637395382,non-hate,0.014278455637395382 +heat,-0.013545539230108261,non-hate,0.013545539230108261 +grave,-0.013527965173125267,non-hate,0.013527965173125267 +keepers,-0.013067572377622128,non-hate,0.013067572377622128 +unhelpful,0.013013465330004692,hate,0.013013465330004692 +tweets,-0.012946579605340958,non-hate,0.012946579605340958 +permissible,-0.012828904204070568,non-hate,0.012828904204070568 +herpes,-0.012761947698891163,non-hate,0.012761947698891163 +upwards,0.012130249291658401,hate,0.012130249291658401 +haters,-0.011888435110449791,non-hate,0.011888435110449791 +twat,0.011619669385254383,hate,0.011619669385254383 +fans,-0.011547603489210209,non-hate,0.011547603489210209 +mammy,0.011262881569564342,hate,0.011262881569564342 +racing,-0.011174283688887954,non-hate,0.011174283688887954 +secret,-0.010837502777576447,non-hate,0.010837502777576447 +plato,0.010055257938802242,hate,0.010055257938802242 +dangerous,0.009965700097382069,hate,0.009965700097382069 +imbeciles,0.00993073359131813,hate,0.00993073359131813 +colored,0.009536509402096272,hate,0.009536509402096272 +propose,-0.009398587979376316,non-hate,0.009398587979376316 +vandalism,0.009304525656625628,hate,0.009304525656625628 +ane,-0.009224571287631989,non-hate,0.009224571287631989 +crisis,0.009179076179862022,hate,0.009179076179862022 +cow,0.009014044888317585,hate,0.009014044888317585 +genetic,-0.008975818753242493,non-hate,0.008975818753242493 +naked,-0.008848296478390694,non-hate,0.008848296478390694 +sixty,0.00878961756825447,hate,0.00878961756825447 +grade,0.00872035976499319,hate,0.00872035976499319 +supermodel,0.008498620241880417,hate,0.008498620241880417 +wade,-0.008491000160574913,non-hate,0.008491000160574913 +shes,0.008426347747445107,hate,0.008426347747445107 +chin,-0.008368161506950855,non-hate,0.008368161506950855 +grid,0.008332611061632633,hate,0.008332611061632633 +es,-0.008152726106345654,non-hate,0.008152726106345654 +unpleasant,-0.008072427473962307,non-hate,0.008072427473962307 diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_english.png b/shap/english_to_hindi_to_hinglish/shap_topwords_english.png new file mode 100644 index 0000000000000000000000000000000000000000..34df8e8ed4769e5b62716efd34548777806d774e --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_english.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cb2eb8c8e31fd8e482186fc42596b14cc031026cb8ee0d49899d4ac5670a3a8 +size 133228 diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_full.csv b/shap/english_to_hindi_to_hinglish/shap_topwords_full.csv new file mode 100644 index 0000000000000000000000000000000000000000..c1b2de3cb07d397aa56daf07f8299f2620a99c6f --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_full.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +rajya,-0.06898477673530579,non-hate,0.06898477673530579 +molvi,0.030005168169736862,hate,0.030005168169736862 +coding,-0.027870049700140953,non-hate,0.027870049700140953 +chalo,0.024187824164982885,hate,0.024187824164982885 +molana,0.023408804088830948,hate,0.023408804088830948 +meat,-0.022356677800416946,non-hate,0.022356677800416946 +haan,-0.02185993455350399,non-hate,0.02185993455350399 +maine,-0.02024389134021476,non-hate,0.02024389134021476 +scn,0.01949775405228138,hate,0.01949775405228138 +elitist,0.018782276660203934,hate,0.018782276660203934 +trophy,0.018481716513633728,hate,0.018481716513633728 +aisi,0.01782621070742607,hate,0.01782621070742607 +lock,0.017817869782447815,hate,0.017817869782447815 +karwa,0.017144904617452994,hate,0.017144904617452994 +kra,-0.01709747686982155,non-hate,0.01709747686982155 +beware,0.01700632832944393,hate,0.01700632832944393 +izzat,-0.01642655022442341,non-hate,0.01642655022442341 +baad,0.01541352178901434,hate,0.01541352178901434 +irish,-0.015048699453473091,non-hate,0.015048699453473091 +tattoo,-0.014861606061458588,non-hate,0.014861606061458588 +reno,-0.014712845906615257,non-hate,0.014712845906615257 +aa,-0.014623998664319515,non-hate,0.014623998664319515 +waly,0.013796273618936539,hate,0.013796273618936539 +chuka,-0.013494111597537994,non-hate,0.013494111597537994 +approximately,0.013476947322487831,hate,0.013476947322487831 +template,-0.013467205091728829,non-hate,0.013467205091728829 +camera,-0.013451387378154323,non-hate,0.013451387378154323 +live,-0.013026186265051365,non-hate,0.013026186265051365 +service,0.012946579605340958,hate,0.012946579605340958 +saab,0.012845651246607304,hate,0.012845651246607304 +svi,0.012820753268897533,hate,0.012820753268897533 +mulla,-0.012600169517099857,non-hate,0.012600169517099857 +sahi,0.012316819978877902,hate,0.012316819978877902 +native,-0.012283542193472385,non-hate,0.012283542193472385 +annoying,0.012117007747292519,hate,0.012117007747292519 +fags,-0.012009791098535061,non-hate,0.012009791098535061 +uh,0.011897200252860785,hate,0.011897200252860785 +online,-0.01150062307715416,non-hate,0.01150062307715416 +hogi,0.011441920883953571,hate,0.011441920883953571 +saadi,0.011334981769323349,hate,0.011334981769323349 +kaise,0.01110095251351595,hate,0.01110095251351595 +phir,0.011077019764343277,hate,0.011077019764343277 +migrants,0.010952260345220566,hate,0.010952260345220566 +dino,0.010864288546144962,hate,0.010864288546144962 +snapchat,0.010446332395076752,hate,0.010446332395076752 +insane,0.010381272993981838,hate,0.010381272993981838 +meets,0.010348992422223091,hate,0.010348992422223091 +wakai,0.009823844768106937,hate,0.009823844768106937 +suna,-0.009709286270663142,non-hate,0.009709286270663142 +bloody,0.009668286889791489,hate,0.009668286889791489 diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_full.png b/shap/english_to_hindi_to_hinglish/shap_topwords_full.png new file mode 100644 index 0000000000000000000000000000000000000000..99f250b979e4c0cfcee510acf818234216d0d0f3 --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_full.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bd1861a63caaa209ad4f314571e5769ae94c30c8f2a21c1557a375d8e1795a0 +size 117892 diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_hindi.csv b/shap/english_to_hindi_to_hinglish/shap_topwords_hindi.csv new file mode 100644 index 0000000000000000000000000000000000000000..49e63a451c90057fbfd950b531969fe03ad345bc --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_hindi.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +आखिरकार,-0.0033358391374349594,non-hate,0.0033358391374349594 +मुख,-0.002794696483761072,non-hate,0.002794696483761072 +बिकुल,-0.002528517274186015,non-hate,0.002528517274186015 +रे,0.0023682015016674995,hate,0.0023682015016674995 +शायद,-0.0022491682320833206,non-hate,0.0022491682320833206 +पढ़ने,0.0022027792874723673,hate,0.0022027792874723673 +ह,-0.002087237313389778,non-hate,0.002087237313389778 +हिसक,-0.0020699563901871443,non-hate,0.0020699563901871443 +लाइव,0.0019216991495341063,hate,0.0019216991495341063 +बताई,-0.0018982397159561515,non-hate,0.0018982397159561515 +सहायता,-0.0018641268834471703,non-hate,0.0018641268834471703 +चमत्कारी,-0.0017948689055629075,non-hate,0.0017948689055629075 +केक,-0.0017150851344922557,non-hate,0.0017150851344922557 +गाला,0.0016695075610186905,hate,0.0016695075610186905 +अपवित्रहोते,-0.0016398112056776881,non-hate,0.0016398112056776881 +आदिवासी,-0.0015912492526695132,non-hate,0.0015912492526695132 +औखत,0.001533194212242961,hate,0.001533194212242961 +जिन,0.001522717298939824,hate,0.001522717298939824 +तुम्हे,-0.0013824340712744743,non-hate,0.0013824340712744743 +थोक,0.0013769678334938362,hate,0.0013769678334938362 +जिंदाबाद,0.0013051220448687673,hate,0.0013051220448687673 +बईमान,-0.0012976985890418291,non-hate,0.0012976985890418291 +रहूंगा,0.0012577249435707927,hate,0.0012577249435707927 +फेसबुक,0.0012535798596218228,hate,0.0012535798596218228 +बजेगी,0.0012123718624934554,hate,0.0012123718624934554 +जूता,-0.0011677799229801167,non-hate,0.0011677799229801167 +दल्ला,0.0011653159781417344,hate,0.0011653159781417344 +चूहोंसे,0.0011476236395537853,hate,0.0011476236395537853 +मादर,0.0011468089651316404,hate,0.0011468089651316404 +यूजर्स,0.0011401970405131578,hate,0.0011401970405131578 +अशक्षित,-0.0011198444990441203,non-hate,0.0011198444990441203 +प्राणी,-0.0011107462923973799,non-hate,0.0011107462923973799 +क्यार,0.0010954404715448618,hate,0.0010954404715448618 +इनके,0.0010914014419540763,hate,0.0010914014419540763 +खुश,-0.0010589224693831056,non-hate,0.0010589224693831056 +चुपचाप,-0.0010541451629251242,non-hate,0.0010541451629251242 +ख,0.0010227515595033765,hate,0.0010227515595033765 +जिहादी,0.0010184063576161861,hate,0.0010184063576161861 +चैंपियन,0.0010129829170182347,hate,0.0010129829170182347 +टिंडर,0.001010704436339438,hate,0.001010704436339438 +बेंच,0.0009955434361472726,hate,0.0009955434361472726 +पुलिसकर्मी,0.000992004992440343,hate,0.000992004992440343 +निशान,0.0009880269644781947,hate,0.0009880269644781947 +चली,0.0009770456235855818,hate,0.0009770456235855818 +कराई,0.0009719990193843842,hate,0.0009719990193843842 +आना,0.0009707278222776949,hate,0.0009707278222776949 +किसके,0.0009676986373960972,hate,0.0009676986373960972 +पलट,0.0009659182978793979,hate,0.0009659182978793979 +प्रशांत,-0.0009649685234762728,non-hate,0.0009649685234762728 +फाइल,0.0009567874949425459,hate,0.0009567874949425459 diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_hindi.png b/shap/english_to_hindi_to_hinglish/shap_topwords_hindi.png new file mode 100644 index 0000000000000000000000000000000000000000..158f0b92ce8a364cd7a49dbf70564b27a18285c0 Binary files /dev/null and b/shap/english_to_hindi_to_hinglish/shap_topwords_hindi.png differ diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.csv b/shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.csv new file mode 100644 index 0000000000000000000000000000000000000000..0beec2c510d6261df349a8ceef0e4a207419eb02 --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +nawaz,0.05676102265715599,hate,0.05676102265715599 +gau,-0.04096746817231178,non-hate,0.04096746817231178 +age,-0.036737654358148575,non-hate,0.036737654358148575 +rajya,-0.03457523917313665,non-hate,0.03457523917313665 +dhawan,0.03389862924814224,hate,0.03389862924814224 +chori,-0.03355357423424721,non-hate,0.03355357423424721 +channels,-0.030348965898156166,non-hate,0.030348965898156166 +bashing,0.025306934490799904,hate,0.025306934490799904 +shareef,0.023180870339274406,hate,0.023180870339274406 +scn,0.022421205416321754,hate,0.022421205416321754 +nd,-0.02009766168581943,non-hate,0.02009766168581943 +dene,-0.019619129442920286,non-hate,0.019619129442920286 +false,0.019098816439509392,hate,0.019098816439509392 +evening,-0.018587205559015274,non-hate,0.018587205559015274 +kra,-0.01842050999403,non-hate,0.01842050999403 +bacchi,0.017619134858250618,hate,0.017619134858250618 +d,-0.01737004518508911,non-hate,0.01737004518508911 +trophy,0.017280252650380135,hate,0.017280252650380135 +income,-0.016550052911043167,non-hate,0.016550052911043167 +faida,-0.016374368220567703,non-hate,0.016374368220567703 +kahne,-0.015819858759641647,non-hate,0.015819858759641647 +pee,0.015452487394213676,hate,0.015452487394213676 +lagu,-0.015273250406607985,non-hate,0.015273250406607985 +leather,0.014458280056715012,hate,0.014458280056715012 +saja,0.014278997123862306,hate,0.014278997123862306 +saab,0.014250840060412884,hate,0.014250840060412884 +deke,-0.014217130839824677,non-hate,0.014217130839824677 +janab,-0.014192691538482904,non-hate,0.014192691538482904 +asha,-0.01403744425624609,non-hate,0.01403744425624609 +bure,0.013829917646944523,hate,0.013829917646944523 +lalu,-0.013798991218209267,non-hate,0.013798991218209267 +khel,-0.013672738336026669,non-hate,0.013672738336026669 +arey,0.013623342849314213,hate,0.013623342849314213 +acha,-0.013561081141233444,non-hate,0.013561081141233444 +zayda,0.013337258249521255,hate,0.013337258249521255 +hafiz,0.01327153667807579,hate,0.01327153667807579 +ruk,0.013265514746308327,hate,0.013265514746308327 +language,-0.013246475718915462,non-hate,0.013246475718915462 +slowly,-0.013072092086076736,non-hate,0.013072092086076736 +chord,-0.013013044372200966,non-hate,0.013013044372200966 +walon,0.012972347748776277,hate,0.012972347748776277 +trolling,-0.012936906889081001,non-hate,0.012936906889081001 +ayesha,-0.012675345875322819,non-hate,0.012675345875322819 +politics,0.01266510970890522,hate,0.01266510970890522 +bache,-0.012600020660708347,non-hate,0.012600020660708347 +baar,0.012325921445153654,hate,0.012325921445153654 +laye,-0.012260080315172672,non-hate,0.012260080315172672 +uh,0.012248085578903556,hate,0.012248085578903556 +sambhal,0.012246502563357353,hate,0.012246502563357353 +piche,0.012035678916921219,hate,0.012035678916921219 diff --git a/shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.png b/shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.png new file mode 100644 index 0000000000000000000000000000000000000000..760beae0c93b3d93b153ab075d2a0df9273b90bf --- /dev/null +++ b/shap/english_to_hindi_to_hinglish/shap_topwords_hinglish.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41bd34ca6211aeed52b86d32ee65d91cc14372e4830a52f18e8acf2a9a6045ae +size 124873 diff --git a/shap/english_to_hinglish_to_hindi/shap_summary.csv b/shap/english_to_hinglish_to_hindi/shap_summary.csv new file mode 100644 index 0000000000000000000000000000000000000000..f0256ac12994419781ccdd3bfe8faa6a79a0f8e7 --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_summary.csv @@ -0,0 +1,5 @@ +model,eval_lang,top_hate_words,top_nonhate_words +english_to_hinglish_to_hindi,english,"grave(0.0338), svi(0.0263), vox(0.0217), ahh(0.0194), grown(0.0186)","buried(-0.0278), coon(-0.0256), ane(-0.0220), million(-0.0207), normally(-0.0188)" +english_to_hinglish_to_hindi,hindi,"नड्डा(0.0055), हिसक(0.0051), बड़े(0.0041), सांसदों(0.0039), रद्दीके(0.0034)","समाज(-0.0014), सबकी(-0.0012), सऊदी(-0.0011), समझा(-0.0008), अमिताभ(-0.0008)" +english_to_hinglish_to_hindi,hinglish,"khi(0.0456), dada(0.0441), kiske(0.0366), srk(0.0360), chalo(0.0347)","gau(-0.0470), tk(-0.0458), online(-0.0412), liberty(-0.0403), taraf(-0.0291)" +english_to_hinglish_to_hindi,full,"dada(0.0407), roj(0.0378), shamelessness(0.0378), tujhe(0.0306), epic(0.0273)","akash(-0.0397), sapna(-0.0320), proud(-0.0257), buddy(-0.0251), episcopal(-0.0195)" diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_english.csv b/shap/english_to_hinglish_to_hindi/shap_topwords_english.csv new file mode 100644 index 0000000000000000000000000000000000000000..36e65ae3fde81b5b2b5aeefa8d61e65ef2c84bea --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_english.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +grave,0.03377414867281914,hate,0.03377414867281914 +buried,-0.02778712473809719,non-hate,0.02778712473809719 +svi,0.026343040333207075,hate,0.026343040333207075 +coon,-0.025583354756236076,non-hate,0.025583354756236076 +ane,-0.022041624411940575,non-hate,0.022041624411940575 +vox,0.021650129929184914,hate,0.021650129929184914 +million,-0.020728571340441704,non-hate,0.020728571340441704 +ahh,0.01938459277153015,hate,0.01938459277153015 +normally,-0.018763625994324684,non-hate,0.018763625994324684 +grown,0.01864521950483322,hate,0.01864521950483322 +arent,0.01821700669825077,hate,0.01821700669825077 +ave,-0.017279399558901787,non-hate,0.017279399558901787 +sahara,0.01677117682993412,hate,0.01677117682993412 +tweets,-0.015860194340348244,non-hate,0.015860194340348244 +meat,0.015826083719730377,hate,0.015826083719730377 +w,-0.015207762829959393,non-hate,0.015207762829959393 +uniform,-0.015109163708984852,non-hate,0.015109163708984852 +hebrew,-0.015015708282589912,non-hate,0.015015708282589912 +balkan,0.014470499940216541,hate,0.014470499940216541 +soldier,-0.014143366133794188,non-hate,0.014143366133794188 +useless,0.014114265330135822,hate,0.014114265330135822 +piss,0.014020537491887808,hate,0.014020537491887808 +pirates,-0.01364984828978777,non-hate,0.01364984828978777 +wwf,0.013570856302976608,hate,0.013570856302976608 +grinning,-0.013469666242599487,non-hate,0.013469666242599487 +secret,-0.012978477403521538,non-hate,0.012978477403521538 +ukrainians,0.012872913852334023,hate,0.012872913852334023 +trans,0.01278695184737444,hate,0.01278695184737444 +beast,-0.012583800591528416,non-hate,0.012583800591528416 +irresponsible,0.012579096481204033,hate,0.012579096481204033 +rema,0.012194445356726646,hate,0.012194445356726646 +lazy,0.012098634615540504,hate,0.012098634615540504 +targeting,0.01180151104927063,hate,0.01180151104927063 +politicians,0.01170935109257698,hate,0.01170935109257698 +thug,-0.01141161285340786,non-hate,0.01141161285340786 +intimidation,0.011395364068448544,hate,0.011395364068448544 +tons,-0.011145942301178971,non-hate,0.011145942301178971 +banning,0.011118147522211075,hate,0.011118147522211075 +digging,-0.011110752820968628,non-hate,0.011110752820968628 +rent,-0.011054765433073044,non-hate,0.011054765433073044 +appalled,0.010620629414916039,hate,0.010620629414916039 +bj,0.010487527586519718,hate,0.010487527586519718 +daily,-0.01038834173232317,non-hate,0.01038834173232317 +eh,0.010360904037952423,hate,0.010360904037952423 +fire,0.010198072530329227,hate,0.010198072530329227 +grows,-0.009938654489815235,non-hate,0.009938654489815235 +os,-0.009923352859914303,non-hate,0.009923352859914303 +vista,-0.009848393954598578,non-hate,0.009848393954598578 +generate,-0.009714319370687008,non-hate,0.009714319370687008 +hacked,-0.009654613211750984,non-hate,0.009654613211750984 diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_english.png b/shap/english_to_hinglish_to_hindi/shap_topwords_english.png new file mode 100644 index 0000000000000000000000000000000000000000..d4b726fbefa375924d818c6204f60c631a044291 --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_english.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:905a144051168a6176fb77c13cf50bba822d3351e20e4002d22d18406513712f +size 123347 diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_full.csv b/shap/english_to_hinglish_to_hindi/shap_topwords_full.csv new file mode 100644 index 0000000000000000000000000000000000000000..360b3da12edfd55718997e8653729114bc5c9ad3 --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_full.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +dada,0.04066853225231171,hate,0.04066853225231171 +akash,-0.039651013910770416,non-hate,0.039651013910770416 +roj,0.03778504952788353,hate,0.03778504952788353 +shamelessness,0.037755079567432404,hate,0.037755079567432404 +sapna,-0.03201965615153313,non-hate,0.03201965615153313 +tujhe,0.030623262748122215,hate,0.030623262748122215 +epic,0.027315154671669006,hate,0.027315154671669006 +inka,0.02694362960755825,hate,0.02694362960755825 +proud,-0.02568676881492138,non-hate,0.02568676881492138 +buddy,-0.025116443634033203,non-hate,0.025116443634033203 +goli,0.02420537918806076,hate,0.02420537918806076 +vox,0.02265656739473343,hate,0.02265656739473343 +tweeter,0.02087932825088501,hate,0.02087932825088501 +episcopal,-0.019459785893559456,non-hate,0.019459785893559456 +fod,0.01903546042740345,hate,0.01903546042740345 +wohi,0.019030367955565453,hate,0.019030367955565453 +monger,0.01893049106001854,hate,0.01893049106001854 +blame,0.018009111285209656,hate,0.018009111285209656 +extremism,0.017395827919244766,hate,0.017395827919244766 +wwf,0.01709640584886074,hate,0.01709640584886074 +coon,-0.017000511288642883,non-hate,0.017000511288642883 +weak,0.016892073675990105,hate,0.016892073675990105 +representing,-0.016812097281217575,non-hate,0.016812097281217575 +server,-0.01660715602338314,non-hate,0.01660715602338314 +spend,-0.01656663417816162,non-hate,0.01656663417816162 +ignorance,0.0164802223443985,hate,0.0164802223443985 +sin,0.016073254868388176,hate,0.016073254868388176 +pappu,-0.015476739034056664,non-hate,0.015476739034056664 +phir,0.014622029470046982,hate,0.014622029470046982 +flop,0.014550437219440937,hate,0.014550437219440937 +baar,0.013954593800008297,hate,0.013954593800008297 +rajput,0.013538122177124023,hate,0.013538122177124023 +milegi,0.013241058215498924,hate,0.013241058215498924 +naam,0.013159419875591993,hate,0.013159419875591993 +typically,-0.013029334135353565,non-hate,0.013029334135353565 +violating,0.01233892235904932,hate,0.01233892235904932 +total,-0.012132490985095501,non-hate,0.012132490985095501 +oops,0.012068760581314564,hate,0.012068760581314564 +merged,-0.012045545503497124,non-hate,0.012045545503497124 +ahh,0.012032249011099339,hate,0.012032249011099339 +pa,-0.011953715234994888,non-hate,0.011953715234994888 +par,-0.011810776049969718,non-hate,0.011810776049969718 +grand,-0.011286125518381596,non-hate,0.011286125518381596 +angela,-0.011194123886525631,non-hate,0.011194123886525631 +golden,0.010730420239269733,hate,0.010730420239269733 +shant,0.010725311934947968,hate,0.010725311934947968 +occurred,0.010573689825832844,hate,0.010573689825832844 +race,0.010471123270690441,hate,0.010471123270690441 +rai,-0.010417665354907513,non-hate,0.010417665354907513 +bachao,-0.010407710447907448,non-hate,0.010407710447907448 diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_full.png b/shap/english_to_hinglish_to_hindi/shap_topwords_full.png new file mode 100644 index 0000000000000000000000000000000000000000..570b30bc9737dd212d92426efd56820c1db18167 --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_full.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:172850fc129d0706db3780825a746e3e0f9c98adcbb4279a1f541956b4ce5fe1 +size 129078 diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_hindi.csv b/shap/english_to_hinglish_to_hindi/shap_topwords_hindi.csv new file mode 100644 index 0000000000000000000000000000000000000000..71a905f7589817d6a20a9d6e80eaf49c9b4d1d3a --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_hindi.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +नड्डा,0.005499556194990873,hate,0.005499556194990873 +हिसक,0.005087146069854498,hate,0.005087146069854498 +बड़े,0.004061257466673851,hate,0.004061257466673851 +सांसदों,0.003906283061951399,hate,0.003906283061951399 +रद्दीके,0.003370868042111397,hate,0.003370868042111397 +क्रिकेट,0.0031474395655095577,hate,0.0031474395655095577 +कथन,0.0029360835906118155,hate,0.0029360835906118155 +मादरचोदो,0.0026648947969079018,hate,0.0026648947969079018 +चाहते,0.002624175696837483,hate,0.002624175696837483 +पसंद,0.002621591789647937,hate,0.002621591789647937 +ड्यूटी,0.002569320611655712,hate,0.002569320611655712 +बोर्ड,0.0025203678283105546,hate,0.0025203678283105546 +हिंसक,0.002390975598245859,hate,0.002390975598245859 +किताब,0.0023411898873746395,hate,0.0023411898873746395 +इस्लाम,0.00226191314868629,hate,0.00226191314868629 +हरामी,0.002196153625845909,hate,0.002196153625845909 +दुखदायी,0.0021855717059224844,hate,0.0021855717059224844 +मरना,0.0021317461505532265,hate,0.0021317461505532265 +कंपनियां,0.0021031014330219477,hate,0.0021031014330219477 +कचरेकेअलावा,0.0020710002863779664,hate,0.0020710002863779664 +मैनेजर,0.0019841075409203768,hate,0.0019841075409203768 +अँगरेज़,0.001956545893335715,hate,0.001956545893335715 +इनकार,0.0019150477601215243,hate,0.0019150477601215243 +सहायता,0.0019010002724826336,hate,0.0019010002724826336 +करेंगे,0.0017668280779616907,hate,0.0017668280779616907 +पीड़ित,0.001677434891462326,hate,0.001677434891462326 +मरेंगे,0.0016757090925239027,hate,0.0016757090925239027 +गर्दन,0.0016647828742861748,hate,0.0016647828742861748 +लिया।,0.0015350208850577474,hate,0.0015350208850577474 +अशक्षित,0.0015324329433497041,hate,0.0015324329433497041 +जानते,0.0015190744306892157,hate,0.0015190744306892157 +भीख,0.0014856359557597898,hate,0.0014856359557597898 +बदलेंगे,0.0014584831660613418,hate,0.0014584831660613418 +जिसको,0.001414740807376802,hate,0.001414740807376802 +निश्चित,0.0014141463907435536,hate,0.0014141463907435536 +फ़ीसदी,0.0013848076341673732,hate,0.0013848076341673732 +समाज,-0.0013823281042277813,non-hate,0.0013823281042277813 +उखाड़,0.0013639141106978059,hate,0.0013639141106978059 +आएगा,0.0013399263843894005,hate,0.0013399263843894005 +सबसे,0.0013189903534415255,hate,0.0013189903534415255 +मेरे,0.0013168124516076724,hate,0.0013168124516076724 +गाला,0.0013123887474648654,hate,0.0013123887474648654 +इतने,0.0013043124347783912,hate,0.0013043124347783912 +जीता,0.0012975430581718683,hate,0.0012975430581718683 +वा,0.0012956272112205625,hate,0.0012956272112205625 +दिमागी,0.0012712612515315413,hate,0.0012712612515315413 +सबकी,-0.0012336482759565115,non-hate,0.0012336482759565115 +उन्हें,0.0012219998966429557,hate,0.0012219998966429557 +आखिरकार,0.0012153195809029664,hate,0.0012153195809029664 +कॉर्पोरेट,0.0011820816434919834,hate,0.0011820816434919834 diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_hindi.png b/shap/english_to_hinglish_to_hindi/shap_topwords_hindi.png new file mode 100644 index 0000000000000000000000000000000000000000..efd8542a02a8ee3c1beaa0e476d6e3bfd62963b3 Binary files /dev/null and b/shap/english_to_hinglish_to_hindi/shap_topwords_hindi.png differ diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.csv b/shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.csv new file mode 100644 index 0000000000000000000000000000000000000000..0d52542a6293166c5e55401d8533d28ffff6b3f1 --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +gau,-0.04699094593524933,non-hate,0.04699094593524933 +tk,-0.04579180106520653,non-hate,0.04579180106520653 +khi,0.0456167496740818,hate,0.0456167496740818 +dada,0.04406077042222023,hate,0.04406077042222023 +online,-0.041158441454172134,non-hate,0.041158441454172134 +liberty,-0.0402672253549099,non-hate,0.0402672253549099 +kiske,0.03656422719359398,hate,0.03656422719359398 +srk,0.036029606402735226,hate,0.036029606402735226 +chalo,0.03472071513533592,hate,0.03472071513533592 +bahana,0.034635912626981735,hate,0.034635912626981735 +sharam,0.03259362338576466,hate,0.03259362338576466 +taraf,-0.02905560121871531,non-hate,0.02905560121871531 +sidh,0.027074351906776428,hate,0.027074351906776428 +crore,-0.026280434802174568,non-hate,0.026280434802174568 +janab,0.025764032965525985,hate,0.025764032965525985 +halat,0.02558947168290615,hate,0.02558947168290615 +kiska,0.024916188791394234,hate,0.024916188791394234 +rajput,0.024207284518827994,hate,0.024207284518827994 +bhut,0.023591337259858847,hate,0.023591337259858847 +kahi,0.023206871984127377,hate,0.023206871984127377 +dino,0.022923404350876808,hate,0.022923404350876808 +camera,-0.022406714037060738,non-hate,0.022406714037060738 +lagu,-0.022218532860279083,non-hate,0.022218532860279083 +dala,0.02199588343501091,hate,0.02199588343501091 +hta,0.021994290873408318,hate,0.021994290873408318 +trolling,-0.021914688870310783,non-hate,0.021914688870310783 +gobar,0.021392202004790306,hate,0.021392202004790306 +gya,0.020908841164782643,hate,0.020908841164782643 +bethe,0.02072063833475113,hate,0.02072063833475113 +tatti,0.020108881639316678,hate,0.020108881639316678 +tweeter,0.020051514729857445,hate,0.020051514729857445 +disgusting,0.020027877762913704,hate,0.020027877762913704 +behan,0.019147705053910613,hate,0.019147705053910613 +code,-0.01852494291961193,non-hate,0.01852494291961193 +bure,0.018249373883008957,hate,0.018249373883008957 +roj,0.017728047711508616,hate,0.017728047711508616 +asim,0.01761671155691147,hate,0.01761671155691147 +insane,0.017560334876179695,hate,0.017560334876179695 +wakai,0.017409587278962135,hate,0.017409587278962135 +geeta,-0.017389291897416115,non-hate,0.017389291897416115 +phone,-0.01737423613667488,non-hate,0.01737423613667488 +replies,-0.01737412065267563,non-hate,0.01737412065267563 +yah,0.017366129904985428,hate,0.017366129904985428 +trophy,0.017336612567305565,hate,0.017336612567305565 +brotherhood,0.017283383756875992,hate,0.017283383756875992 +dose,-0.017220083624124527,non-hate,0.017220083624124527 +papa,0.01703502455105384,hate,0.01703502455105384 +khel,-0.016657598316669464,non-hate,0.016657598316669464 +azeem,0.016594119369983673,hate,0.016594119369983673 +load,-0.01650962419807911,non-hate,0.01650962419807911 diff --git a/shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.png b/shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.png new file mode 100644 index 0000000000000000000000000000000000000000..cf57db7e2e41d63b36089b5816eb0b9a10b268f9 --- /dev/null +++ b/shap/english_to_hinglish_to_hindi/shap_topwords_hinglish.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a73eaa1335c649f903c7a64de6a7740d5e722ba74fcc19554ee67ff7443ef74a +size 116991 diff --git a/shap/hindi_to_english_to_hinglish/shap_summary.csv b/shap/hindi_to_english_to_hinglish/shap_summary.csv new file mode 100644 index 0000000000000000000000000000000000000000..c304fb4859c67d3eb52589349a72276b7ef03607 --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_summary.csv @@ -0,0 +1,5 @@ +model,eval_lang,top_hate_words,top_nonhate_words +hindi_to_english_to_hinglish,english,"credence(0.0465), bj(0.0348), rosario(0.0261), ghazi(0.0247), eni(0.0242)","plain(-0.0263), stranger(-0.0251), sarcasm(-0.0244), rubbish(-0.0195), comprise(-0.0192)" +hindi_to_english_to_hinglish,hindi,"कॉल(0.0046), भूमिपूजन(0.0041), लें(0.0037), आधी(0.0033), मूर्ख(0.0031)","मैसेज(-0.0021), बेमन(-0.0013), पुलिसकर्मी(-0.0012), जाएगी(-0.0012), पड़े(-0.0012)" +hindi_to_english_to_hinglish,hinglish,"bacchi(0.0463), bull(0.0451), srk(0.0271), bahana(0.0246), behan(0.0244)","madrassa(-0.0449), zaida(-0.0328), gdp(-0.0294), bech(-0.0257), nd(-0.0252)" +hindi_to_english_to_hinglish,full,"skua(0.0816), brut(0.0428), cleansing(0.0313), captaincy(0.0308), baar(0.0264)","taraf(-0.0303), pussy(-0.0279), directory(-0.0232), quran(-0.0224), kaha(-0.0183)" diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_english.csv b/shap/hindi_to_english_to_hinglish/shap_topwords_english.csv new file mode 100644 index 0000000000000000000000000000000000000000..f2f3948a586b3a8a34fb4c39589cd68223081c95 --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_english.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +credence,0.04645003750920296,hate,0.04645003750920296 +bj,0.034841448068618774,hate,0.034841448068618774 +plain,-0.026305072009563446,non-hate,0.026305072009563446 +rosario,0.026062248274683952,hate,0.026062248274683952 +stranger,-0.025110220536589622,non-hate,0.025110220536589622 +ghazi,0.024667367339134216,hate,0.024667367339134216 +sarcasm,-0.024418074637651443,non-hate,0.024418074637651443 +eni,0.024185189977288246,hate,0.024185189977288246 +original,0.023739179650268245,hate,0.023739179650268245 +purpose,0.022877716802759095,hate,0.022877716802759095 +meat,0.02180168218910694,hate,0.02180168218910694 +rubbish,-0.01947150006890297,non-hate,0.01947150006890297 +comprise,-0.019205201417207718,non-hate,0.019205201417207718 +fiasco,0.016540812328457832,hate,0.016540812328457832 +sahara,0.015753297445674736,hate,0.015753297445674736 +vandalism,0.015700376010499895,hate,0.015700376010499895 +cooking,-0.01567102037370205,non-hate,0.01567102037370205 +funds,-0.01552804559469223,non-hate,0.01552804559469223 +bloody,0.014914470724761486,hate,0.014914470724761486 +revealing,-0.014907026663422585,non-hate,0.014907026663422585 +definition,-0.013764521107077599,non-hate,0.013764521107077599 +circus,0.013064487837255001,hate,0.013064487837255001 +accusations,0.012445050291717052,hate,0.012445050291717052 +village,-0.012163950843387283,non-hate,0.012163950843387283 +criticize,-0.012133165262639523,non-hate,0.012133165262639523 +mere,-0.012098387815058231,non-hate,0.012098387815058231 +mike,0.012091495096683502,hate,0.012091495096683502 +salad,-0.012074373662471771,non-hate,0.012074373662471771 +multicultural,0.011976183392107487,hate,0.011976183392107487 +2017,-0.011603245511651039,non-hate,0.011603245511651039 +remarks,0.011279447004199028,hate,0.011279447004199028 +arguments,0.011224827729165554,hate,0.011224827729165554 +shamanic,-0.01111504528671503,non-hate,0.01111504528671503 +coding,-0.011073308065533638,non-hate,0.011073308065533638 +lunatics,-0.011048647575080395,non-hate,0.011048647575080395 +choke,0.010911933146417141,hate,0.010911933146417141 +glove,0.010900042951107025,hate,0.010900042951107025 +responsible,0.010841659270226955,hate,0.010841659270226955 +volition,-0.010667622089385986,non-hate,0.010667622089385986 +irrelevant,0.010649058036506176,hate,0.010649058036506176 +radical,0.010174911469221115,hate,0.010174911469221115 +wikimedia,-0.010040879249572754,non-hate,0.010040879249572754 +advantageously,-0.01004065852612257,non-hate,0.01004065852612257 +arrogance,0.010037285275757313,hate,0.010037285275757313 +registered,-0.010005852207541466,non-hate,0.010005852207541466 +kicking,0.009971213527023792,hate,0.009971213527023792 +behaving,0.009872743859887123,hate,0.009872743859887123 +its,-0.009824934541029506,non-hate,0.009824934541029506 +stance,0.009769201278686523,hate,0.009769201278686523 +smack,-0.009650851949118078,non-hate,0.009650851949118078 diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_english.png b/shap/hindi_to_english_to_hinglish/shap_topwords_english.png new file mode 100644 index 0000000000000000000000000000000000000000..03eda66a01cd191b878ce6fd75ef655daf66d6a5 --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_english.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d188d588ecd20d78673bd3dc04805a1ffc61eaf7514629a944dd8d5427b4614d +size 136987 diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_full.csv b/shap/hindi_to_english_to_hinglish/shap_topwords_full.csv new file mode 100644 index 0000000000000000000000000000000000000000..8f059ea6a220415e91260df4f66bb8ce04c1a41f --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_full.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +skua,0.08158993721008301,hate,0.08158993721008301 +brut,0.04280721768736839,hate,0.04280721768736839 +cleansing,0.03134329244494438,hate,0.03134329244494438 +captaincy,0.030838793143630028,hate,0.030838793143630028 +taraf,-0.030284931883215904,non-hate,0.030284931883215904 +pussy,-0.02786024659872055,non-hate,0.02786024659872055 +baar,0.02642027847468853,hate,0.02642027847468853 +kasai,0.025917520746588707,hate,0.025917520746588707 +molvi,0.02466745860874653,hate,0.02466745860874653 +directory,-0.023185718804597855,non-hate,0.023185718804597855 +quran,-0.02236177772283554,non-hate,0.02236177772283554 +captain,0.022211698815226555,hate,0.022211698815226555 +sail,0.021913528442382812,hate,0.021913528442382812 +declared,0.02165551856160164,hate,0.02165551856160164 +bhool,0.020629791542887688,hate,0.020629791542887688 +walon,0.0198285523802042,hate,0.0198285523802042 +kaha,-0.018295519131546218,non-hate,0.018295519131546218 +aman,-0.017780931666493416,non-hate,0.017780931666493416 +machinations,0.01660303771495819,hate,0.01660303771495819 +lete,0.016594698652625084,hate,0.016594698652625084 +coding,-0.01642676256597042,non-hate,0.01642676256597042 +lathi,0.016386056318879128,hate,0.016386056318879128 +lagu,-0.016071297228336334,non-hate,0.016071297228336334 +abb,0.0160201508551836,hate,0.0160201508551836 +janab,0.016011323779821396,hate,0.016011323779821396 +turkey,0.015641644597053528,hate,0.015641644597053528 +hu,-0.014945299364626408,non-hate,0.014945299364626408 +kashmir,0.014508838765323162,hate,0.014508838765323162 +aata,0.01447764877229929,hate,0.01447764877229929 +poor,0.014436762779951096,hate,0.014436762779951096 +subtle,-0.014343751594424248,non-hate,0.014343751594424248 +dekh,0.01430919673293829,hate,0.01430919673293829 +attack,-0.014237020397558808,non-hate,0.014237020397558808 +khi,0.014189592562615871,hate,0.014189592562615871 +pr,0.014058244414627552,hate,0.014058244414627552 +avoid,0.013930351473391056,hate,0.013930351473391056 +agr,0.013830370580156645,hate,0.013830370580156645 +karna,0.013334315037354827,hate,0.013334315037354827 +chalo,0.013232632540166378,hate,0.013232632540166378 +mulla,-0.01312780100852251,non-hate,0.01312780100852251 +spic,0.01289000641554594,hate,0.01289000641554594 +ragini,-0.012712415307760239,non-hate,0.012712415307760239 +jin,-0.01269172690808773,non-hate,0.01269172690808773 +remark,0.01267898827791214,hate,0.01267898827791214 +bare,-0.012546997517347336,non-hate,0.012546997517347336 +khana,0.01244534645229578,hate,0.01244534645229578 +bashing,0.012364937923848629,hate,0.012364937923848629 +khuda,0.012339971028268337,hate,0.012339971028268337 +bloody,0.012339216656982899,hate,0.012339216656982899 +trolling,-0.012257399347921213,non-hate,0.012257399347921213 diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_full.png b/shap/hindi_to_english_to_hinglish/shap_topwords_full.png new file mode 100644 index 0000000000000000000000000000000000000000..ebcdd94bdae8f6d8aaf7d1f802432b4ce20c36a9 --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_full.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13fb869c87b3e10482b348e8fab110fcc2857ac3a8770a0447ab22f13a6cc24f +size 128775 diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_hindi.csv b/shap/hindi_to_english_to_hinglish/shap_topwords_hindi.csv new file mode 100644 index 0000000000000000000000000000000000000000..339a95d665435aab0362cb5cd33fb4406d383875 --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_hindi.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +कॉल,0.004648116417229176,hate,0.004648116417229176 +भूमिपूजन,0.00412707356736064,hate,0.00412707356736064 +लें,0.0037075933068990707,hate,0.0037075933068990707 +आधी,0.003285990096628666,hate,0.003285990096628666 +मूर्ख,0.0031396488193422556,hate,0.0031396488193422556 +क,0.002917179372161627,hate,0.002917179372161627 +राजीव,0.0027316459454596043,hate,0.0027316459454596043 +पीकर,0.0025358640123158693,hate,0.0025358640123158693 +काल्पनिक,0.0022344260942190886,hate,0.0022344260942190886 +करदिया,0.0022163160610944033,hate,0.0022163160610944033 +सदियों,0.0021872264333069324,hate,0.0021872264333069324 +खालसा,0.0021593221463263035,hate,0.0021593221463263035 +मैसेज,-0.0021256704349070787,non-hate,0.0021256704349070787 +पत्रकारिता,0.002100299810990691,hate,0.002100299810990691 +निर्वस्त्र,0.0020729624666273594,hate,0.0020729624666273594 +खाऊंगा,0.0020205227192491293,hate,0.0020205227192491293 +दी।,0.0020052604377269745,hate,0.0020052604377269745 +पर्यावरण,0.001958520384505391,hate,0.001958520384505391 +खोलकर,0.0019040468614548445,hate,0.0019040468614548445 +नज़दीकी,0.0018469050992280245,hate,0.0018469050992280245 +सचिवालय,0.0018418339313939214,hate,0.0018418339313939214 +अध्यक्ष,0.0018362035043537617,hate,0.0018362035043537617 +भीड़,0.001832223730161786,hate,0.001832223730161786 +आसपास,0.0018116296268999577,hate,0.0018116296268999577 +प्रकाश,0.0017764391377568245,hate,0.0017764391377568245 +नारा,0.0017635419499129057,hate,0.0017635419499129057 +स्थापित,0.0017599757993593812,hate,0.0017599757993593812 +मी,0.001748561393469572,hate,0.001748561393469572 +अबे,0.0017446019919589162,hate,0.0017446019919589162 +सीखा,0.001734658144414425,hate,0.001734658144414425 +चिड़ियाघर,0.0017311526462435722,hate,0.0017311526462435722 +वेद,0.0017284646164625883,hate,0.0017284646164625883 +अगली,0.0017243522452190518,hate,0.0017243522452190518 +कहने,0.0016829832457005978,hate,0.0016829832457005978 +बाप,0.001677687163464725,hate,0.001677687163464725 +नियम,0.0016645262949168682,hate,0.0016645262949168682 +कचरे,0.0016632884507998824,hate,0.0016632884507998824 +जेल,0.0016590725863352418,hate,0.0016590725863352418 +ऑफिस,0.0016274387016892433,hate,0.0016274387016892433 +जिसे,0.0016175805358216166,hate,0.0016175805358216166 +चाहूंगा,0.0015872376970946789,hate,0.0015872376970946789 +निकाला,0.001551219029352069,hate,0.001551219029352069 +मज़ाक,0.0015492744278162718,hate,0.0015492744278162718 +गांड,0.0015414789086207747,hate,0.0015414789086207747 +जगत,0.001535353367216885,hate,0.001535353367216885 +अवैध,0.0015317134093493223,hate,0.0015317134093493223 +रेपिस्ट,0.0015180136542767286,hate,0.0015180136542767286 +याचिका,0.0015106289647519588,hate,0.0015106289647519588 +आख़िरकार,0.0014834972098469734,hate,0.0014834972098469734 +ग्रेटर,0.0014711980475112796,hate,0.0014711980475112796 diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_hindi.png b/shap/hindi_to_english_to_hinglish/shap_topwords_hindi.png new file mode 100644 index 0000000000000000000000000000000000000000..3b74e834f22a948eec764ee13ca044245d820cb6 Binary files /dev/null and b/shap/hindi_to_english_to_hinglish/shap_topwords_hindi.png differ diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.csv b/shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.csv new file mode 100644 index 0000000000000000000000000000000000000000..a0b07e12d9bc1aba9395e995d6ba19362553ca41 --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +bacchi,0.046268280595541,hate,0.046268280595541 +bull,0.04512413591146469,hate,0.04512413591146469 +madrassa,-0.04489169642329216,non-hate,0.04489169642329216 +zaida,-0.032846979796886444,non-hate,0.032846979796886444 +gdp,-0.029350487515330315,non-hate,0.029350487515330315 +srk,0.027143995370715858,hate,0.027143995370715858 +bech,-0.025672174990177155,non-hate,0.025672174990177155 +nd,-0.02520766109228134,non-hate,0.02520766109228134 +azam,-0.02500207908451557,non-hate,0.02500207908451557 +bahana,0.024554820731282234,hate,0.024554820731282234 +behan,0.024406446143984795,hate,0.024406446143984795 +tatti,0.023972761817276478,hate,0.023972761817276478 +janab,0.023465278893127106,hate,0.023465278893127106 +pok,-0.02334669977426529,non-hate,0.02334669977426529 +rajya,-0.02303268387913704,non-hate,0.02303268387913704 +molvi,0.02216806448996067,hate,0.02216806448996067 +income,-0.021810831502079964,non-hate,0.021810831502079964 +ik,-0.021652765572071075,non-hate,0.021652765572071075 +nahe,0.02129768766462803,hate,0.02129768766462803 +e,-0.02061024119029753,non-hate,0.02061024119029753 +72,-0.020374976564198732,non-hate,0.020374976564198732 +siva,-0.020169060677289963,non-hate,0.020169060677289963 +fod,0.019986752420663834,hate,0.019986752420663834 +every,-0.019902456551790237,non-hate,0.019902456551790237 +khi,0.019523052498698235,hate,0.019523052498698235 +hy,0.019500816240906715,hate,0.019500816240906715 +nahin,0.017115247435867786,hate,0.017115247435867786 +vi,0.01685914397239685,hate,0.01685914397239685 +90,-0.016503230358163517,non-hate,0.016503230358163517 +yup,-0.01625201292335987,non-hate,0.01625201292335987 +leti,0.016210559755563736,hate,0.016210559755563736 +ani,0.015870854258537292,hate,0.015870854258537292 +ur,-0.015786100178956985,non-hate,0.015786100178956985 +age,-0.0156858346114556,non-hate,0.0156858346114556 +dur,-0.015627197921276093,non-hate,0.015627197921276093 +harami,0.015580840408802032,hate,0.015580840408802032 +arey,0.0153880063444376,hate,0.0153880063444376 +sonia,0.01523885503411293,hate,0.01523885503411293 +te,-0.01520167663693428,non-hate,0.01520167663693428 +music,-0.015192155726253986,non-hate,0.015192155726253986 +du,-0.015182935865595937,non-hate,0.015182935865595937 +ganda,-0.015164026990532875,non-hate,0.015164026990532875 +lagu,-0.01504000648856163,non-hate,0.01504000648856163 +sharam,0.014934451901353896,hate,0.014934451901353896 +aao,0.014927598589565605,hate,0.014927598589565605 +dala,0.014842050150036812,hate,0.014842050150036812 +famous,0.014752278104424477,hate,0.014752278104424477 +stop,0.014676655642688274,hate,0.014676655642688274 +sister,-0.014637755850950876,non-hate,0.014637755850950876 +paye,-0.01454240083694458,non-hate,0.01454240083694458 diff --git a/shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.png b/shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.png new file mode 100644 index 0000000000000000000000000000000000000000..ef093b20d4abf0348552838b1030f9ff4becbe3f --- /dev/null +++ b/shap/hindi_to_english_to_hinglish/shap_topwords_hinglish.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d1faa850c49647326c73b6d6e8c7f696318d2afc8f843df89a4b0436e67fb78 +size 116339 diff --git a/shap/hindi_to_hinglish_to_english/shap_summary.csv b/shap/hindi_to_hinglish_to_english/shap_summary.csv new file mode 100644 index 0000000000000000000000000000000000000000..cd46a150058db463f7abaa395aaf476aaecc9de1 --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_summary.csv @@ -0,0 +1,5 @@ +model,eval_lang,top_hate_words,top_nonhate_words +hindi_to_hinglish_to_english,english,"violence(0.0719), ahh(0.0715), potus(0.0362), spic(0.0294), undocumented(0.0293)","beginner(-0.0293), dollars(-0.0248), bih(-0.0239), messages(-0.0226), total(-0.0225)" +hindi_to_hinglish_to_english,hindi,"जाएगी(0.0056), दूसरों(0.0036), इंटरव्यू(0.0036), हवाई(0.0035), अक्षय(0.0035)","नियम(-0.0022), डाला।ये(-0.0022), दर्शन(-0.0020), मुलाकात(-0.0014), उज्ज्वल(-0.0013)" +hindi_to_hinglish_to_english,hinglish,"tatti(0.0287), sham(0.0242), dino(0.0240), roko(0.0207), krk(0.0206)","lac(-0.0421), online(-0.0238), ancestor(-0.0191), zaida(-0.0187), target(-0.0174)" +hindi_to_hinglish_to_english,full,"ahh(0.0703), moi(0.0637), bj(0.0437), fault(0.0353), pan(0.0353)","asperger(-0.0459), hundred(-0.0333), database(-0.0256), wicked(-0.0227), nam(-0.0209)" diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_english.csv b/shap/hindi_to_hinglish_to_english/shap_topwords_english.csv new file mode 100644 index 0000000000000000000000000000000000000000..5092982ea40bf4b62cb9eece07bb07566518e248 --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_english.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +violence,0.0718970075249672,hate,0.0718970075249672 +ahh,0.07150574028491974,hate,0.07150574028491974 +potus,0.0362129770219326,hate,0.0362129770219326 +spic,0.02940574660897255,hate,0.02940574660897255 +beginner,-0.029332537204027176,non-hate,0.029332537204027176 +undocumented,0.029292572289705276,hate,0.029292572289705276 +dollars,-0.02477841079235077,non-hate,0.02477841079235077 +bih,-0.023915961384773254,non-hate,0.023915961384773254 +slant,0.02339380979537964,hate,0.02339380979537964 +lock,0.02322426252067089,hate,0.02322426252067089 +chris,0.02274729311466217,hate,0.02274729311466217 +messages,-0.02258477360010147,non-hate,0.02258477360010147 +total,-0.022455638274550438,non-hate,0.022455638274550438 +pussy,-0.020164035260677338,non-hate,0.020164035260677338 +spaz,0.02013159543275833,hate,0.02013159543275833 +restarted,0.020001359283924103,hate,0.020001359283924103 +rudy,0.01964670605957508,hate,0.01964670605957508 +exhaust,0.018840793520212173,hate,0.018840793520212173 +roughly,-0.0187721848487854,non-hate,0.0187721848487854 +ruin,0.01839724462479353,hate,0.01839724462479353 +reversed,0.01828077994287014,hate,0.01828077994287014 +mica,-0.018118800595402718,non-hate,0.018118800595402718 +gate,0.017711204942315817,hate,0.017711204942315817 +wales,-0.017481878399848938,non-hate,0.017481878399848938 +lebron,-0.017470847815275192,non-hate,0.017470847815275192 +wishes,-0.017086025327444077,non-hate,0.017086025327444077 +per,-0.016760966973379254,non-hate,0.016760966973379254 +ass,0.01571407536539482,hate,0.01571407536539482 +unfounded,0.015698591247200966,hate,0.015698591247200966 +self,0.01537333196029067,hate,0.01537333196029067 +handle,-0.015088985208421946,non-hate,0.015088985208421946 +eliminating,0.014800118139646656,hate,0.014800118139646656 +unacceptable,0.014394650235772133,hate,0.014394650235772133 +vandalism,0.013978655226916695,hate,0.013978655226916695 +lesbos,0.013956660404801369,hate,0.013956660404801369 +garbage,-0.01382022948625187,non-hate,0.01382022948625187 +don,-0.013604654232040048,non-hate,0.013604654232040048 +pedophiles,-0.013143415562808514,non-hate,0.013143415562808514 +query,0.012974048033356667,hate,0.012974048033356667 +croatian,0.012909981422126293,hate,0.012909981422126293 +bored,-0.01279514143243432,non-hate,0.01279514143243432 +backwards,-0.01274916809052229,non-hate,0.01274916809052229 +biceps,0.01270436029881239,hate,0.01270436029881239 +pathetic,0.012687018141150475,hate,0.012687018141150475 +sc,-0.012650532647967339,non-hate,0.012650532647967339 +mars,-0.012649756856262684,non-hate,0.012649756856262684 +failure,0.012612069956958294,hate,0.012612069956958294 +graduated,-0.012397725135087967,non-hate,0.012397725135087967 +domination,0.01227420661598444,hate,0.01227420661598444 +fucked,0.012226648318270842,hate,0.012226648318270842 diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_english.png b/shap/hindi_to_hinglish_to_english/shap_topwords_english.png new file mode 100644 index 0000000000000000000000000000000000000000..5e4702c636f2913230c2223e73dbe380caa3d52c --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_english.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7d98465cc11800c1aec64609586baa6f9f6ecbf493be38cee717a7938c25b99 +size 131513 diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_full.csv b/shap/hindi_to_hinglish_to_english/shap_topwords_full.csv new file mode 100644 index 0000000000000000000000000000000000000000..61fc234ff0371373aa703f0400a30f975941221d --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_full.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +ahh,0.07030271738767624,hate,0.07030271738767624 +moi,0.06365898996591568,hate,0.06365898996591568 +asperger,-0.04588022828102112,non-hate,0.04588022828102112 +bj,0.043733712285757065,hate,0.043733712285757065 +fault,0.03531529754400253,hate,0.03531529754400253 +pan,0.03527500480413437,hate,0.03527500480413437 +hundred,-0.03330186381936073,non-hate,0.03330186381936073 +civilian,0.031162988394498825,hate,0.031162988394498825 +database,-0.025607790797948837,non-hate,0.025607790797948837 +shamelessness,0.025555744767189026,hate,0.025555744767189026 +exhaust,0.024258642457425594,hate,0.024258642457425594 +wicked,-0.022660793736577034,non-hate,0.022660793736577034 +spilling,0.021830838173627853,hate,0.021830838173627853 +mar,0.021807485725730658,hate,0.021807485725730658 +sathi,0.021615521982312202,hate,0.021615521982312202 +overdose,0.02130974270403385,hate,0.02130974270403385 +nam,-0.020943190902471542,non-hate,0.020943190902471542 +shut,0.02044683663795392,hate,0.02044683663795392 +gaya,-0.02016141265630722,non-hate,0.02016141265630722 +stood,-0.019478697329759598,non-hate,0.019478697329759598 +kaisa,0.018417688086628914,hate,0.018417688086628914 +census,-0.017953457310795784,non-hate,0.017953457310795784 +smirking,0.01778489537537098,hate,0.01778489537537098 +thoda,0.01712045632302761,hate,0.01712045632302761 +exists,-0.01634456031024456,non-hate,0.01634456031024456 +ja,0.01629845192655921,hate,0.01629845192655921 +sleepy,-0.016062110662460327,non-hate,0.016062110662460327 +graying,-0.01598961278796196,non-hate,0.01598961278796196 +carter,0.01547007542103529,hate,0.01547007542103529 +allele,-0.015424293465912342,non-hate,0.015424293465912342 +rajput,0.015396442264318466,hate,0.015396442264318466 +handle,-0.014967868104577065,non-hate,0.014967868104577065 +aww,0.0148957259953022,hate,0.0148957259953022 +tvo,0.0148436538875103,hate,0.0148436538875103 +krk,0.014596456661820412,hate,0.014596456661820412 +shoot,-0.01442080270498991,non-hate,0.01442080270498991 +sirf,0.014394848607480526,hate,0.014394848607480526 +apna,-0.014066954026930034,non-hate,0.014066954026930034 +devil,-0.013987665995955467,non-hate,0.013987665995955467 +girls,-0.013944365084171295,non-hate,0.013944365084171295 +fat,-0.013877103105187416,non-hate,0.013877103105187416 +density,-0.013781795278191566,non-hate,0.013781795278191566 +somewhere,-0.013780158013105392,non-hate,0.013780158013105392 +miserable,0.013712558895349503,hate,0.013712558895349503 +zero,-0.01370886666700244,non-hate,0.01370886666700244 +palestine,0.013645797967910767,hate,0.013645797967910767 +gag,0.013614475727081299,hate,0.013614475727081299 +speak,-0.013539474457502365,non-hate,0.013539474457502365 +misogynists,-0.013521498069167137,non-hate,0.013521498069167137 +criticism,0.013326193206012249,hate,0.013326193206012249 diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_full.png b/shap/hindi_to_hinglish_to_english/shap_topwords_full.png new file mode 100644 index 0000000000000000000000000000000000000000..b3f639d58b7630df0914975841c40db136b4830e --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_full.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb9a50cdfd81632370b2f117ba746a62b2ad8abe891a1b6f7e54808ce92ba31b +size 130385 diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_hindi.csv b/shap/hindi_to_hinglish_to_english/shap_topwords_hindi.csv new file mode 100644 index 0000000000000000000000000000000000000000..50eaeb8d5f09707d2845fb3a06ddf599a2a327b8 --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_hindi.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +जाएगी,0.0056337942369282246,hate,0.0056337942369282246 +दूसरों,0.00360527983866632,hate,0.00360527983866632 +इंटरव्यू,0.003584775608032942,hate,0.003584775608032942 +हवाई,0.003547343658283353,hate,0.003547343658283353 +अक्षय,0.003511772956699133,hate,0.003511772956699133 +फुर्सत,0.003444105386734009,hate,0.003444105386734009 +प्रमुख,0.0032255209516733885,hate,0.0032255209516733885 +इजाजत,0.00312454579398036,hate,0.00312454579398036 +बागी,0.0031167282722890377,hate,0.0031167282722890377 +शुभ,0.0031011952087283134,hate,0.0031011952087283134 +लिया।,0.0030351057648658752,hate,0.0030351057648658752 +दिखेंगे,0.0030304554384201765,hate,0.0030304554384201765 +मारूंगा,0.002880593528971076,hate,0.002880593528971076 +आने,0.0028464117098110364,hate,0.0028464117098110364 +चुटिया,0.002836276136804372,hate,0.002836276136804372 +बदलाव,0.002800874412059784,hate,0.002800874412059784 +मिली,0.0027539189904928207,hate,0.0027539189904928207 +गिटार,0.0027533237589523196,hate,0.0027533237589523196 +जुड़े,0.002690871013328433,hate,0.002690871013328433 +त्वरित,0.002655700547620654,hate,0.002655700547620654 +पोस्टमॉर्टम,0.00265315268188715,hate,0.00265315268188715 +साझा,0.0025513144209980965,hate,0.0025513144209980965 +गांड,0.002507714554667473,hate,0.002507714554667473 +फिलहाल,0.002471733605489135,hate,0.002471733605489135 +पेट्रोल,0.0024703029775992036,hate,0.0024703029775992036 +घोषणा,0.00246704020537436,hate,0.00246704020537436 +माफ,0.00243789900559932,hate,0.00243789900559932 +डूबना,0.0022771492367610335,hate,0.0022771492367610335 +बढ़कर,0.002260180888697505,hate,0.002260180888697505 +रखेगा,0.0022215174976736307,hate,0.0022215174976736307 +ज़िंदा,0.0022190422751009464,hate,0.0022190422751009464 +बांटते,0.002208010759204626,hate,0.002208010759204626 +नियम,-0.0022036260925233364,non-hate,0.0022036260925233364 +वकील,0.0021956597920507193,hate,0.0021956597920507193 +अगली,0.0021683110389858484,hate,0.0021683110389858484 +स्टैंड,0.002167156431823969,hate,0.002167156431823969 +डाला।ये,-0.0021572429686784744,non-hate,0.0021572429686784744 +आसपास,0.0021200485061854124,hate,0.0021200485061854124 +आदत,0.0021145101927686483,hate,0.0021145101927686483 +चाहूंगा,0.002094015246257186,hate,0.002094015246257186 +पीड़ित,0.0020374981686472893,hate,0.0020374981686472893 +दर्शन,-0.002007972914725542,non-hate,0.002007972914725542 +सावधानी,0.002003937726840377,hate,0.002003937726840377 +संशोधन,0.001988169038668275,hate,0.001988169038668275 +बनाते,0.0019814681727439165,hate,0.0019814681727439165 +सफाई,0.001927752047777176,hate,0.001927752047777176 +हकीकत,0.0019255629740655422,hate,0.0019255629740655422 +बिलकुल,0.0019199085072614253,hate,0.0019199085072614253 +लगाते,0.0019128392450511456,hate,0.0019128392450511456 +बताते,0.001898040878586471,hate,0.001898040878586471 diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_hindi.png b/shap/hindi_to_hinglish_to_english/shap_topwords_hindi.png new file mode 100644 index 0000000000000000000000000000000000000000..30d2e621820498fa6157cb2a7876a24fd4af3200 Binary files /dev/null and b/shap/hindi_to_hinglish_to_english/shap_topwords_hindi.png differ diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.csv b/shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.csv new file mode 100644 index 0000000000000000000000000000000000000000..0dcb6a3bfefd47b37c58a4e86d6f4f687d36cc35 --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +lac,-0.042069509625434875,non-hate,0.042069509625434875 +tatti,0.028744549956172705,hate,0.028744549956172705 +sham,0.024232318624854088,hate,0.024232318624854088 +dino,0.02402765117585659,hate,0.02402765117585659 +online,-0.023813826963305473,non-hate,0.023813826963305473 +roko,0.020704982802271843,hate,0.020704982802271843 +krk,0.020585810765624046,hate,0.020585810765624046 +lete,0.020226562046445906,hate,0.020226562046445906 +blame,0.019940845668315887,hate,0.019940845668315887 +bure,0.01964375376701355,hate,0.01964375376701355 +ancestor,-0.019133463501930237,non-hate,0.019133463501930237 +zaida,-0.018675070255994797,non-hate,0.018675070255994797 +stop,0.01843524786333243,hate,0.01843524786333243 +cow,0.017804839958747227,hate,0.017804839958747227 +save,0.017410658998414874,hate,0.017410658998414874 +target,-0.017365742940455675,non-hate,0.017365742940455675 +rss,-0.017073381692171097,non-hate,0.017073381692171097 +gau,-0.017007868736982346,non-hate,0.017007868736982346 +kch,0.016833083238452673,hate,0.016833083238452673 +101,-0.01678640954196453,non-hate,0.01678640954196453 +trophy,0.0159439779818058,hate,0.0159439779818058 +madrassa,-0.01594107784330845,non-hate,0.01594107784330845 +rhi,0.015927101537171338,hate,0.015927101537171338 +chori,0.01588687114417553,hate,0.01588687114417553 +muslims,0.015709418104961514,hate,0.015709418104961514 +ra,0.015227079391479492,hate,0.015227079391479492 +tune,-0.015014691278338432,non-hate,0.015014691278338432 +vikash,0.015004333574324846,hate,0.015004333574324846 +baki,0.014981868298491463,hate,0.014981868298491463 +service,-0.014977164566516876,non-hate,0.014977164566516876 +dunia,0.014679802348837256,hate,0.014679802348837256 +politicians,0.01451462134718895,hate,0.01451462134718895 +bacchi,0.014466872438788414,hate,0.014466872438788414 +lo,0.01444881958886981,hate,0.01444881958886981 +sth,0.014444015920162201,hate,0.014444015920162201 +yar,0.014425559621304273,hate,0.014425559621304273 +lathi,0.014414900913834572,hate,0.014414900913834572 +jae,-0.014343447983264923,non-hate,0.014343447983264923 +q,0.014333151979371905,hate,0.014333151979371905 +lada,-0.014195562340319157,non-hate,0.014195562340319157 +condemn,0.014155733399093151,hate,0.014155733399093151 +rakshas,0.014124168083071709,hate,0.014124168083071709 +false,0.014104630798101425,hate,0.014104630798101425 +bharat,-0.013742012855469842,non-hate,0.013742012855469842 +aya,-0.013234891928732395,non-hate,0.013234891928732395 +banane,0.013135356828570366,hate,0.013135356828570366 +parr,-0.013027128763496876,non-hate,0.013027128763496876 +jolly,-0.013003858737647533,non-hate,0.013003858737647533 +chalo,0.0128032939392142,hate,0.0128032939392142 +per,-0.012776409403886645,non-hate,0.012776409403886645 diff --git a/shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.png b/shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.png new file mode 100644 index 0000000000000000000000000000000000000000..ab3d5ca5e556b9d6d9833c3199a09fdf8441aa3e --- /dev/null +++ b/shap/hindi_to_hinglish_to_english/shap_topwords_hinglish.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c97c73f16e5ec679e19f2bdded4fa6361cfaadc835590e4d63b94ce6022508d9 +size 117826 diff --git a/shap/hinglish_to_english_to_hindi/shap_summary.csv b/shap/hinglish_to_english_to_hindi/shap_summary.csv new file mode 100644 index 0000000000000000000000000000000000000000..9dcee5d84e088bb60f7c2aacd60c6bdd079109c0 --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_summary.csv @@ -0,0 +1,5 @@ +model,eval_lang,top_hate_words,top_nonhate_words +hinglish_to_english_to_hindi,english,"bastard(0.0342), establishes(0.0337), code(0.0279), poo(0.0220), hub(0.0218)","blatantly(-0.0509), languages(-0.0296), turkey(-0.0282), fags(-0.0219), gear(-0.0197)" +hinglish_to_english_to_hindi,hindi,"रंजन(0.0049), गोगोई(0.0040), नड्डा(0.0036), सांसदों(0.0036), पित्त(0.0033)","चूतिए(-0.0020), मुल्ले(-0.0019), हथियार(-0.0018), उपनिषद(-0.0015), जन्मभूमि(-0.0014)" +hinglish_to_english_to_hindi,hinglish,"huye(0.0608), dada(0.0565), abb(0.0500), arey(0.0500), abduction(0.0471)","rajya(-0.0759), bahu(-0.0492), parliament(-0.0388), code(-0.0371), music(-0.0339)" +hinglish_to_english_to_hindi,full,"skua(0.0452), praised(0.0339), spic(0.0303), sabse(0.0301), plz(0.0284)","liberty(-0.0376), languages(-0.0262), speaks(-0.0241), bache(-0.0228), maine(-0.0201)" diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_english.csv b/shap/hinglish_to_english_to_hindi/shap_topwords_english.csv new file mode 100644 index 0000000000000000000000000000000000000000..7000b58e184be841628fee9ebc789f2b5da89360 --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_english.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +blatantly,-0.05089934170246124,non-hate,0.05089934170246124 +bastard,0.03424622863531113,hate,0.03424622863531113 +establishes,0.033674389123916626,hate,0.033674389123916626 +languages,-0.02962074801325798,non-hate,0.02962074801325798 +turkey,-0.028210505843162537,non-hate,0.028210505843162537 +code,0.02794194407761097,hate,0.02794194407761097 +poo,0.021984202787280083,hate,0.021984202787280083 +fags,-0.02188652753829956,non-hate,0.02188652753829956 +hub,0.02177145704627037,hate,0.02177145704627037 +grown,0.021145431324839592,hate,0.021145431324839592 +gear,-0.01969330944120884,non-hate,0.01969330944120884 +native,-0.01952183060348034,non-hate,0.01952183060348034 +strikebreaker,-0.01889657787978649,non-hate,0.01889657787978649 +undocumented,0.018055077642202377,hate,0.018055077642202377 +independent,-0.017914675176143646,non-hate,0.017914675176143646 +anne,-0.017657000571489334,non-hate,0.017657000571489334 +conversation,-0.01578935608267784,non-hate,0.01578935608267784 +greatness,0.015463314950466156,hate,0.015463314950466156 +mobile,-0.01528094056993723,non-hate,0.01528094056993723 +breathing,-0.015064464882016182,non-hate,0.015064464882016182 +battleground,-0.015015914104878902,non-hate,0.015015914104878902 +million,0.014449150301516056,hate,0.014449150301516056 +sneak,-0.013950327411293983,non-hate,0.013950327411293983 +wake,-0.01367312390357256,non-hate,0.01367312390357256 +speaks,-0.013496821746230125,non-hate,0.013496821746230125 +rename,-0.013368615880608559,non-hate,0.013368615880608559 +cheers,0.013196121901273727,hate,0.013196121901273727 +dakota,-0.012742486782371998,non-hate,0.012742486782371998 +camber,0.012729320675134659,hate,0.012729320675134659 +crook,-0.012579857371747494,non-hate,0.012579857371747494 +ane,-0.012288169004023075,non-hate,0.012288169004023075 +reputable,-0.012238690629601479,non-hate,0.012238690629601479 +beware,0.012046086601912975,hate,0.012046086601912975 +president,0.011942957527935505,hate,0.011942957527935505 +bih,-0.011916973628103733,non-hate,0.011916973628103733 +violence,0.011849150527268648,hate,0.011849150527268648 +tastes,-0.01170691754668951,non-hate,0.01170691754668951 +captain,-0.01157486904412508,non-hate,0.01157486904412508 +behold,-0.011561742052435875,non-hate,0.011561742052435875 +bosnia,-0.011406704550608993,non-hate,0.011406704550608993 +nt,-0.011282934807240963,non-hate,0.011282934807240963 +redirect,-0.011155341751873493,non-hate,0.011155341751873493 +presidents,0.011127662844955921,hate,0.011127662844955921 +killing,0.010869013145565987,hate,0.010869013145565987 +per,-0.010833741282112896,non-hate,0.010833741282112896 +dude,-0.010654857611370971,non-hate,0.010654857611370971 +desert,0.010481158271431923,hate,0.010481158271431923 +exaggerating,-0.010432315990328789,non-hate,0.010432315990328789 +bloody,0.010380396464218697,hate,0.010380396464218697 +fits,0.010372105985879898,hate,0.010372105985879898 diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_english.png b/shap/hinglish_to_english_to_hindi/shap_topwords_english.png new file mode 100644 index 0000000000000000000000000000000000000000..651397bce9b1f5516071bde7fabc3970f029910d --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_english.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90e0d64515a2e115d6fd96730b55a7d5c1573c198e5fed2cc70f3a000b2b94ba +size 147763 diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_full.csv b/shap/hinglish_to_english_to_hindi/shap_topwords_full.csv new file mode 100644 index 0000000000000000000000000000000000000000..d551752b42bc3864679061b18159ba094770ccd1 --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_full.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +skua,0.045165300369262695,hate,0.045165300369262695 +liberty,-0.03758999705314636,non-hate,0.03758999705314636 +praised,0.03389711678028107,hate,0.03389711678028107 +spic,0.030283592641353607,hate,0.030283592641353607 +sabse,0.030072808265686035,hate,0.030072808265686035 +plz,0.02842707186937332,hate,0.02842707186937332 +languages,-0.02621418423950672,non-hate,0.02621418423950672 +speaks,-0.02408372238278389,non-hate,0.02408372238278389 +hero,0.023479520343244076,hate,0.023479520343244076 +bache,-0.02283983863890171,non-hate,0.02283983863890171 +neta,0.021335992962121964,hate,0.021335992962121964 +nomination,0.021331384778022766,hate,0.021331384778022766 +shilpa,0.021195484790951014,hate,0.021195484790951014 +dig,0.02013213373720646,hate,0.02013213373720646 +maine,-0.020081011578440666,non-hate,0.020081011578440666 +purely,-0.019505105912685394,non-hate,0.019505105912685394 +strikebreaker,-0.019220121204853058,non-hate,0.019220121204853058 +fee,-0.019008541479706764,non-hate,0.019008541479706764 +phone,0.018852755427360535,hate,0.018852755427360535 +underneath,-0.017952201887965202,non-hate,0.017952201887965202 +lada,-0.017737718299031258,non-hate,0.017737718299031258 +gi,-0.017614235325405996,non-hate,0.017614235325405996 +condemn,0.017232690006494522,hate,0.017232690006494522 +rapists,-0.016899345442652702,non-hate,0.016899345442652702 +encourage,0.016003252938389778,hate,0.016003252938389778 +hm,-0.015780076384544373,non-hate,0.015780076384544373 +giri,-0.015388287603855133,non-hate,0.015388287603855133 +huye,0.015341192483901978,hate,0.015341192483901978 +chali,0.015011332929134369,hate,0.015011332929134369 +sharm,0.014891122467815876,hate,0.014891122467815876 +phir,-0.014537704410031438,non-hate,0.014537704410031438 +alien,-0.014496441930532455,non-hate,0.014496441930532455 +conversation,-0.01447097584605217,non-hate,0.01447097584605217 +dharm,0.014261177740991116,hate,0.014261177740991116 +chori,0.014171980321407318,hate,0.014171980321407318 +musalman,0.013522947207093239,hate,0.013522947207093239 +karna,0.013492814032360911,hate,0.013492814032360911 +ghr,0.013368877582252026,hate,0.013368877582252026 +banna,0.013349328190088272,hate,0.013349328190088272 +banana,0.013253219425678253,hate,0.013253219425678253 +potus,0.012928418815135956,hate,0.012928418815135956 +due,0.012706820853054523,hate,0.012706820853054523 +wiping,0.012546641752123833,hate,0.012546641752123833 +zaida,-0.012538600713014603,non-hate,0.012538600713014603 +tab,-0.012485527588675419,non-hate,0.012485527588675419 +discriminatory,0.012484968639910221,hate,0.012484968639910221 +farmers,-0.012337804771959782,non-hate,0.012337804771959782 +hone,-0.012320831359829754,non-hate,0.012320831359829754 +doubt,0.012252646498382092,hate,0.012252646498382092 +rocks,0.012190154753625393,hate,0.012190154753625393 diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_full.png b/shap/hinglish_to_english_to_hindi/shap_topwords_full.png new file mode 100644 index 0000000000000000000000000000000000000000..e7250dd9735eda08d5477ddd4503479c2eb319a5 --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_full.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db2d607dc8260dcc41c5c64b32777341da961d8737b3ffa4315fe302dbdd66b2 +size 131619 diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_hindi.csv b/shap/hinglish_to_english_to_hindi/shap_topwords_hindi.csv new file mode 100644 index 0000000000000000000000000000000000000000..db1975e51735fa8e290dd40698d6798e5a0f6af0 --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_hindi.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +रंजन,0.004858044907450676,hate,0.004858044907450676 +गोगोई,0.004045760724693537,hate,0.004045760724693537 +नड्डा,0.003624022239819169,hate,0.003624022239819169 +सांसदों,0.00355382333509624,hate,0.00355382333509624 +पित्त,0.003283682744950056,hate,0.003283682744950056 +जस्टिस,0.0032806815579533577,hate,0.0032806815579533577 +मानवाधिकार,0.003065404947847128,hate,0.003065404947847128 +चाहते,0.0026524588465690613,hate,0.0026524588465690613 +मुख,0.002507484285160899,hate,0.002507484285160899 +नाही,0.00243033142760396,hate,0.00243033142760396 +तुम्हे,0.0023580568959005177,hate,0.0023580568959005177 +इनकार,0.0022621112875640392,hate,0.0022621112875640392 +डेड,0.0021442491561174393,hate,0.0021442491561174393 +विचारधारा,0.0021222170908004045,hate,0.0021222170908004045 +चूतिए,-0.0020024559926241636,non-hate,0.0020024559926241636 +चूतिये,0.0019985511316917837,hate,0.0019985511316917837 +गे,0.0019334255484864116,hate,0.0019334255484864116 +मुल्ले,-0.0019252224592491984,non-hate,0.0019252224592491984 +रोहित,0.0018914320971816778,hate,0.0018914320971816778 +कश्मीरी,0.0018896303517976776,hate,0.0018896303517976776 +चूहों,0.0018517603166401386,hate,0.0018517603166401386 +हथियार,-0.0018309983424842358,non-hate,0.0018309983424842358 +कमीने,0.0017851601587608457,hate,0.0017851601587608457 +सीईओ,0.0017522264970466495,hate,0.0017522264970466495 +वेद,0.0017456603236496449,hate,0.0017456603236496449 +जेएनयू,0.0017385744722560048,hate,0.0017385744722560048 +सजा,0.0017214178806170821,hate,0.0017214178806170821 +कियु,0.0017036357894539833,hate,0.0017036357894539833 +पुराण,0.0016952299047261477,hate,0.0016952299047261477 +बॉडी,0.001682272064499557,hate,0.001682272064499557 +कबड़,0.0016750880749896169,hate,0.0016750880749896169 +हु,0.001672996673732996,hate,0.001672996673732996 +अबे,0.0016666783703840338,hate,0.0016666783703840338 +अपवित्रहोते,0.0016425950452685356,hate,0.0016425950452685356 +कचरे,0.001636024215258658,hate,0.001636024215258658 +छुडाप,0.0016270331107079983,hate,0.0016270331107079983 +बेशक,0.001598592847585678,hate,0.001598592847585678 +इम्युनिटी,0.001598526258021593,hate,0.001598526258021593 +केटीएम,0.0015730351442471147,hate,0.0015730351442471147 +स्थिति,0.0015697261551395059,hate,0.0015697261551395059 +मक्कार,0.0015637140022590756,hate,0.0015637140022590756 +जायेगा,0.001558025018312037,hate,0.001558025018312037 +गद्दार,0.0015458859270438552,hate,0.0015458859270438552 +अर्पित,0.001543360180221498,hate,0.001543360180221498 +बईमान,0.0015231017605401576,hate,0.0015231017605401576 +जला,0.0015192392747849226,hate,0.0015192392747849226 +रखनी,0.0015158919850364327,hate,0.0015158919850364327 +उपनिषद,-0.0015126250218600035,non-hate,0.0015126250218600035 +निसान,0.0014855031622573733,hate,0.0014855031622573733 +जन्मभूमि,-0.0014218830037862062,non-hate,0.0014218830037862062 diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_hindi.png b/shap/hinglish_to_english_to_hindi/shap_topwords_hindi.png new file mode 100644 index 0000000000000000000000000000000000000000..2204648fd12fadf6c4cfd44b6946fd7a5f8f8f8c Binary files /dev/null and b/shap/hinglish_to_english_to_hindi/shap_topwords_hindi.png differ diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.csv b/shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.csv new file mode 100644 index 0000000000000000000000000000000000000000..f0560435ce43a4716517888ac1068c07c96db750 --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +rajya,-0.0759279802441597,non-hate,0.0759279802441597 +huye,0.06079550087451935,hate,0.06079550087451935 +dada,0.056532178074121475,hate,0.056532178074121475 +abb,0.050049543380737305,hate,0.050049543380737305 +arey,0.049972862005233765,hate,0.049972862005233765 +bahu,-0.049183763563632965,non-hate,0.049183763563632965 +abduction,0.04710543155670166,hate,0.04710543155670166 +saab,0.04519256576895714,hate,0.04519256576895714 +parliament,-0.03884896636009216,non-hate,0.03884896636009216 +code,-0.03713630512356758,non-hate,0.03713630512356758 +ghr,0.0367225781083107,hate,0.0367225781083107 +fod,0.034826572984457016,hate,0.034826572984457016 +music,-0.033922940492630005,non-hate,0.033922940492630005 +specific,-0.033881500363349915,non-hate,0.033881500363349915 +karny,0.0332486592233181,hate,0.0332486592233181 +age,-0.03195729851722717,non-hate,0.03195729851722717 +mardon,0.029221178963780403,hate,0.029221178963780403 +kapil,0.029139980673789978,hate,0.029139980673789978 +azam,-0.028090525418519974,non-hate,0.028090525418519974 +nawaz,-0.027591099962592125,non-hate,0.027591099962592125 +liberty,-0.027396081015467644,non-hate,0.027396081015467644 +if,-0.026691650971770287,non-hate,0.026691650971770287 +neta,0.026557128876447678,hate,0.026557128876447678 +slowly,-0.026449978351593018,non-hate,0.026449978351593018 +khi,0.0263580484315753,hate,0.0263580484315753 +poor,0.02515023946762085,hate,0.02515023946762085 +nari,0.02403189428150654,hate,0.02403189428150654 +asim,0.02378745749592781,hate,0.02378745749592781 +haram,0.023602023720741272,hate,0.023602023720741272 +punjab,0.02333788387477398,hate,0.02333788387477398 +channels,-0.023088950663805008,non-hate,0.023088950663805008 +messages,-0.02251121588051319,non-hate,0.02251121588051319 +area,-0.022255538031458855,non-hate,0.022255538031458855 +behan,0.022177519276738167,hate,0.022177519276738167 +10th,0.02112431824207306,hate,0.02112431824207306 +bure,0.021056117489933968,hate,0.021056117489933968 +acid,-0.021044211462140083,non-hate,0.021044211462140083 +service,-0.020889309234917164,non-hate,0.020889309234917164 +krk,0.020752448588609695,hate,0.020752448588609695 +72,-0.020707347110146657,non-hate,0.020707347110146657 +jase,0.02063090354204178,hate,0.02063090354204178 +save,0.020494550932198763,hate,0.020494550932198763 +monger,0.020482024177908897,hate,0.020482024177908897 +rhi,0.0203238669433631,hate,0.0203238669433631 +agr,0.02029761633214851,hate,0.02029761633214851 +theory,-0.020191151648759842,non-hate,0.020191151648759842 +per,-0.019894008824944888,non-hate,0.019894008824944888 +lo,0.019664086850493083,hate,0.019664086850493083 +marke,0.019314537290483713,hate,0.019314537290483713 +blasts,0.019097572192549706,hate,0.019097572192549706 diff --git a/shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.png b/shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.png new file mode 100644 index 0000000000000000000000000000000000000000..f9a220ebb8f6027e9f116f99c720f21a4f3f7bba --- /dev/null +++ b/shap/hinglish_to_english_to_hindi/shap_topwords_hinglish.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87813dec81453b42ec9af8a4fa2f421edbd23a4eebe982ea6df1234ddae0ec6f +size 128394 diff --git a/shap/hinglish_to_hindi_to_english/shap_summary.csv b/shap/hinglish_to_hindi_to_english/shap_summary.csv new file mode 100644 index 0000000000000000000000000000000000000000..649cb1a4d63f46774fd52a4d1f960b70d51f6c15 --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_summary.csv @@ -0,0 +1,5 @@ +model,eval_lang,top_hate_words,top_nonhate_words +hinglish_to_hindi_to_english,english,"opponents(0.0395), massacres(0.0308), coon(0.0305), ahh(0.0301), fitness(0.0297)","annie(-0.0499), model(-0.0371), nearly(-0.0243), lloyd(-0.0235), nest(-0.0221)" +hinglish_to_hindi_to_english,hindi,"लें(0.0086), अमिताभ(0.0058), मी(0.0056), करेंगे(0.0047), रखता(0.0047)","आंखें(-0.0045), जे(-0.0044), बुरे(-0.0019), लोगे(-0.0017), जायज़ा(-0.0016)" +hinglish_to_hindi_to_english,hinglish,"fav(0.0607), janab(0.0521), chori(0.0496), cum(0.0447), ruk(0.0444)","online(-0.0643), gau(-0.0528), dehli(-0.0524), 2017(-0.0523), rajya(-0.0422)" +hinglish_to_hindi_to_english,full,"srk(0.0681), roj(0.0613), rhi(0.0345), purana(0.0345), aapke(0.0303)","nest(-0.0303), maine(-0.0298), hone(-0.0239), haired(-0.0194), barrel(-0.0173)" diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_english.csv b/shap/hinglish_to_hindi_to_english/shap_topwords_english.csv new file mode 100644 index 0000000000000000000000000000000000000000..fc7a6251d2ea37d722fb87d1d0228f040abbfb65 --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_english.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +annie,-0.04990576580166817,non-hate,0.04990576580166817 +opponents,0.03946081921458244,hate,0.03946081921458244 +model,-0.03713728487491608,non-hate,0.03713728487491608 +massacres,0.030780743807554245,hate,0.030780743807554245 +coon,0.030544811859726906,hate,0.030544811859726906 +ahh,0.0300755575299263,hate,0.0300755575299263 +fitness,0.029730938374996185,hate,0.029730938374996185 +andre,0.029647817835211754,hate,0.029647817835211754 +nearly,-0.024262990802526474,non-hate,0.024262990802526474 +lloyd,-0.023539507761597633,non-hate,0.023539507761597633 +nest,-0.02208160236477852,non-hate,0.02208160236477852 +balkan,0.02194832358509302,hate,0.02194832358509302 +denounce,0.02125898189842701,hate,0.02125898189842701 +bloody,0.020301640033721924,hate,0.020301640033721924 +rings,-0.019803544506430626,non-hate,0.019803544506430626 +ism,0.01783180423080921,hate,0.01783180423080921 +cretin,-0.01731354556977749,non-hate,0.01731354556977749 +killer,-0.016857897862792015,non-hate,0.016857897862792015 +messing,-0.016541702672839165,non-hate,0.016541702672839165 +piss,0.0156863396987319,hate,0.0156863396987319 +ck,-0.0149026894941926,non-hate,0.0149026894941926 +homosexuality,0.014809366315603256,hate,0.014809366315603256 +hearts,-0.014587335288524628,non-hate,0.014587335288524628 +versus,-0.014346254043630324,non-hate,0.014346254043630324 +diego,0.014238875359296799,hate,0.014238875359296799 +persecution,0.014087439514696598,hate,0.014087439514696598 +kad,0.014003830030560493,hate,0.014003830030560493 +thick,-0.013994825072586536,non-hate,0.013994825072586536 +snow,-0.013510975055396557,non-hate,0.013510975055396557 +ignorance,0.013386040925979614,hate,0.013386040925979614 +declared,0.013379760086536407,hate,0.013379760086536407 +navigate,-0.013187539763748646,non-hate,0.013187539763748646 +struggling,0.0131745720282197,hate,0.0131745720282197 +barack,0.012917174026370049,hate,0.012917174026370049 +relationships,-0.012719886377453804,non-hate,0.012719886377453804 +causing,-0.012439743615686893,non-hate,0.012439743615686893 +pathetic,0.012179793207906187,hate,0.012179793207906187 +messages,-0.011999010108411312,non-hate,0.011999010108411312 +shut,0.011349649226758629,hate,0.011349649226758629 +girlfriend,-0.010944552719593048,non-hate,0.010944552719593048 +perica,0.010894902050495148,hate,0.010894902050495148 +chemical,-0.010857944376766682,non-hate,0.010857944376766682 +medina,0.010757159441709518,hate,0.010757159441709518 +turns,-0.010626721195876598,non-hate,0.010626721195876598 +ignoring,0.010389825329184532,hate,0.010389825329184532 +cleansing,0.010249228216707706,hate,0.010249228216707706 +tragedies,0.01022716797888279,hate,0.01022716797888279 +seal,0.010033071972429752,hate,0.010033071972429752 +concise,0.009975466877222061,hate,0.009975466877222061 +unable,-0.00986519642174244,non-hate,0.00986519642174244 diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_english.png b/shap/hinglish_to_hindi_to_english/shap_topwords_english.png new file mode 100644 index 0000000000000000000000000000000000000000..91f806f944a6caedd42ca0cc9632f82e9586aa7c --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_english.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42fd8696438c757891480cacb93b732916250bff6799fc5c881c2dd3df547ffa +size 128053 diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_full.csv b/shap/hinglish_to_hindi_to_english/shap_topwords_full.csv new file mode 100644 index 0000000000000000000000000000000000000000..eb3710099385e5646d8caf44f44cd960f6509013 --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_full.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +srk,0.06805661972612143,hate,0.06805661972612143 +roj,0.06128497049212456,hate,0.06128497049212456 +rhi,0.034514395985752344,hate,0.034514395985752344 +purana,0.034482844173908234,hate,0.034482844173908234 +aapke,0.030325796455144882,hate,0.030325796455144882 +nest,-0.030280647799372673,non-hate,0.030280647799372673 +maine,-0.02983572706580162,non-hate,0.02983572706580162 +neta,0.02925512008368969,hate,0.02925512008368969 +sambhal,0.028608132153749466,hate,0.028608132153749466 +accused,0.02803703024983406,hate,0.02803703024983406 +poove,0.02556428872048855,hate,0.02556428872048855 +hone,-0.023881315253674984,non-hate,0.023881315253674984 +banane,0.022616036236286163,hate,0.022616036236286163 +bhaag,0.021927908062934875,hate,0.021927908062934875 +minsk,0.021603794768452644,hate,0.021603794768452644 +taras,0.021583624184131622,hate,0.021583624184131622 +peene,0.020969515666365623,hate,0.020969515666365623 +kal,0.020435119047760963,hate,0.020435119047760963 +haired,-0.01935916393995285,non-hate,0.01935916393995285 +raat,0.019197316840291023,hate,0.019197316840291023 +dishonest,0.019153328612446785,hate,0.019153328612446785 +lesbos,0.017389073967933655,hate,0.017389073967933655 +barrel,-0.01730499230325222,non-hate,0.01730499230325222 +sach,0.01712094433605671,hate,0.01712094433605671 +curse,0.017062122002243996,hate,0.017062122002243996 +choke,0.017054671421647072,hate,0.017054671421647072 +isko,-0.016832545399665833,non-hate,0.016832545399665833 +nahin,0.0167633518576622,hate,0.0167633518576622 +humain,0.01674545556306839,hate,0.01674545556306839 +object,-0.016262399032711983,non-hate,0.016262399032711983 +janta,-0.01621467061340809,non-hate,0.01621467061340809 +bibi,0.016138317063450813,hate,0.016138317063450813 +hell,0.015529275871813297,hate,0.015529275871813297 +saja,0.015517775900661945,hate,0.015517775900661945 +dene,-0.01533447578549385,non-hate,0.01533447578549385 +bugs,-0.015224321745336056,non-hate,0.015224321745336056 +kare,-0.014794052578508854,non-hate,0.014794052578508854 +foul,0.014439523220062256,hate,0.014439523220062256 +mante,0.014361357316374779,hate,0.014361357316374779 +kisi,-0.0140520337348183,non-hate,0.0140520337348183 +govt,-0.013569456525146961,non-hate,0.013569456525146961 +idiotic,0.01343502663075924,hate,0.01343502663075924 +stop,0.013344064083260795,hate,0.013344064083260795 +apni,0.013016708195209503,hate,0.013016708195209503 +hame,0.012874514795839787,hate,0.012874514795839787 +kahin,0.012834189459681511,hate,0.012834189459681511 +fag,0.012802124954760075,hate,0.012802124954760075 +zaida,-0.012758812867105007,non-hate,0.012758812867105007 +kaam,0.012605081079527736,hate,0.012605081079527736 +hu,-0.01245044730603695,non-hate,0.01245044730603695 diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_full.png b/shap/hinglish_to_hindi_to_english/shap_topwords_full.png new file mode 100644 index 0000000000000000000000000000000000000000..d4f1662b5ce1a5f13f0c9fe994bcb2cdfc809d38 --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_full.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b504b63a1e5ba968a5879d7f81c4cf2b66fc3a3edf14a7b65d361586fa3f0377 +size 119956 diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_hindi.csv b/shap/hinglish_to_hindi_to_english/shap_topwords_hindi.csv new file mode 100644 index 0000000000000000000000000000000000000000..f6bc60ec4ec2ec961a849c23031bb03c93d03dfa --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_hindi.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +लें,0.008626047521829605,hate,0.008626047521829605 +अमिताभ,0.005800061859190464,hate,0.005800061859190464 +मी,0.005557241849601269,hate,0.005557241849601269 +करेंगे,0.004731479100883007,hate,0.004731479100883007 +रखता,0.004669889574870467,hate,0.004669889574870467 +कचरेकेअलावा,0.004584053531289101,hate,0.004584053531289101 +आंखें,-0.004464127589017153,non-hate,0.004464127589017153 +जे,-0.004386733286082745,non-hate,0.004386733286082745 +बुद्धिमान,0.004032979719340801,hate,0.004032979719340801 +तुम्हे,0.003640879178419709,hate,0.003640879178419709 +कॉल,0.003331782529130578,hate,0.003331782529130578 +नड्डा,0.0031585940159857273,hate,0.0031585940159857273 +मायने,0.003097150765825063,hate,0.003097150765825063 +खोलकर,0.003090438898652792,hate,0.003090438898652792 +जाते,0.0030073935631662607,hate,0.0030073935631662607 +मादरचोदो,0.0029459020006470382,hate,0.0029459020006470382 +बड़े,0.0029280518647283316,hate,0.0029280518647283316 +निफल,0.0027771680615842342,hate,0.0027771680615842342 +वास्तव,0.00268734572455287,hate,0.00268734572455287 +रे,0.0025570588186383247,hate,0.0025570588186383247 +कहता,0.002113583730533719,hate,0.002113583730533719 +इमोजी,0.00210297922603786,hate,0.00210297922603786 +अपवाद,0.0020661079324781895,hate,0.0020661079324781895 +बढ़,0.0020095540676265955,hate,0.0020095540676265955 +बुरे,-0.0019087349064648151,non-hate,0.0019087349064648151 +तुन,0.0019037906313315034,hate,0.0019037906313315034 +हत्यारे,0.0018699637148529291,hate,0.0018699637148529291 +दुखद,0.0018564447527751327,hate,0.0018564447527751327 +खड़ूस,0.0017999258707277477,hate,0.0017999258707277477 +करवाई,0.0017876203637570143,hate,0.0017876203637570143 +गिटार,0.001748509588651359,hate,0.001748509588651359 +ताजा,0.0017347505781799555,hate,0.0017347505781799555 +लोगे,-0.001715033664368093,non-hate,0.001715033664368093 +हिंसक,0.001694944512564689,hate,0.001694944512564689 +और,0.0016802815593301172,hate,0.0016802815593301172 +वेटर,0.001663697068579495,hate,0.001663697068579495 +गा,0.001641901326365769,hate,0.001641901326365769 +जस्टिस,0.0016388087533414364,hate,0.0016388087533414364 +हम्म,0.0016045832016970962,hate,0.0016045832016970962 +जायज़ा,-0.0015765810385346413,non-hate,0.0015765810385346413 +खुलने,0.0015732976607978344,hate,0.0015732976607978344 +सब,0.0015729534256024636,hate,0.0015729534256024636 +मुर्गी,0.0015594522701576352,hate,0.0015594522701576352 +ऑफिस,0.001547174877487123,hate,0.001547174877487123 +देती,0.0015463022282347083,hate,0.0015463022282347083 +गांड,0.0015205388359997112,hate,0.0015205388359997112 +छत,-0.0014913232298567891,non-hate,0.0014913232298567891 +लाइव,0.0014834352768957615,hate,0.0014834352768957615 +प्रशासन,0.0014470192836597562,hate,0.0014470192836597562 +सांसदों,0.0014441311359405518,hate,0.0014441311359405518 diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_hindi.png b/shap/hinglish_to_hindi_to_english/shap_topwords_hindi.png new file mode 100644 index 0000000000000000000000000000000000000000..b27d5fb84b690ca672402dd3134a4c67fc927a31 Binary files /dev/null and b/shap/hinglish_to_hindi_to_english/shap_topwords_hindi.png differ diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.csv b/shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.csv new file mode 100644 index 0000000000000000000000000000000000000000..3e5e0c9cbb90a9cef17d9bf11e1c7edd1cf61f47 --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.csv @@ -0,0 +1,51 @@ +word,mean_shap,direction,abs_shap +online,-0.06433086842298508,non-hate,0.06433086842298508 +fav,0.06065862253308296,hate,0.06065862253308296 +gau,-0.0528433695435524,non-hate,0.0528433695435524 +dehli,-0.05239486321806908,non-hate,0.05239486321806908 +2017,-0.052268192172050476,non-hate,0.052268192172050476 +janab,0.05212605558335781,hate,0.05212605558335781 +chori,0.04961038753390312,hate,0.04961038753390312 +cum,0.04466735199093819,hate,0.04466735199093819 +ruk,0.044430702924728394,hate,0.044430702924728394 +arey,0.042845964431762695,hate,0.042845964431762695 +rajya,-0.042195793241262436,non-hate,0.042195793241262436 +dada,0.04133421555161476,hate,0.04133421555161476 +molesting,-0.04054604098200798,non-hate,0.04054604098200798 +archives,-0.03537296503782272,non-hate,0.03537296503782272 +karwa,0.034824684262275696,hate,0.034824684262275696 +bahu,-0.033206261694431305,non-hate,0.033206261694431305 +dene,-0.03283716691657901,non-hate,0.03283716691657901 +roko,0.032627955079078674,hate,0.032627955079078674 +haram,0.031285516917705536,hate,0.031285516917705536 +baar,0.029772340133786202,hate,0.029772340133786202 +rapists,-0.028908738866448402,non-hate,0.028908738866448402 +suar,0.02859204200406869,hate,0.02859204200406869 +sakti,-0.028141846880316734,non-hate,0.028141846880316734 +nara,0.027445178478956223,hate,0.027445178478956223 +marke,0.027013015002012253,hate,0.027013015002012253 +purana,0.026843277737498283,hate,0.026843277737498283 +bda,0.026482844725251198,hate,0.026482844725251198 +pedophile,-0.026333538815379143,non-hate,0.026333538815379143 +dafa,-0.026259953156113625,non-hate,0.026259953156113625 +aya,-0.026245612651109695,non-hate,0.026245612651109695 +bashing,0.02614543028175831,hate,0.02614543028175831 +khabar,-0.026133114472031593,non-hate,0.026133114472031593 +karny,0.025935150682926178,hate,0.025935150682926178 +torture,-0.02555154077708721,non-hate,0.02555154077708721 +cow,0.024551453068852425,hate,0.024551453068852425 +sharam,0.02370516350492835,hate,0.02370516350492835 +khel,-0.023591392673552036,non-hate,0.023591392673552036 +lock,0.023504426702857018,hate,0.023504426702857018 +rhi,0.0230955658480525,hate,0.0230955658480525 +bull,0.02281995862722397,hate,0.02281995862722397 +srk,0.022796668065711855,hate,0.022796668065711855 +bahana,0.0227511003613472,hate,0.0227511003613472 +sal,0.02244492620229721,hate,0.02244492620229721 +chord,-0.022226957604289055,non-hate,0.022226957604289055 +age,-0.022157187496001523,non-hate,0.022157187496001523 +gi,-0.02194811621059974,non-hate,0.02194811621059974 +save,0.021887936163693666,hate,0.021887936163693666 +chaudhary,0.021629609167575836,hate,0.021629609167575836 +tatti,0.02162465127184987,hate,0.02162465127184987 +entry,0.021456779912114143,hate,0.021456779912114143 diff --git a/shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.png b/shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.png new file mode 100644 index 0000000000000000000000000000000000000000..6a8aae1ff695145dc4cb228fbb7a7dc0fcda0cc5 --- /dev/null +++ b/shap/hinglish_to_hindi_to_english/shap_topwords_hinglish.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db6db65d4160b9d7f8c720a38e0a498b9c82f725a020139fe1d9d5ce3e9631dc +size 121751