post_id
int64
text
string
xlmr
float64
xlmt
float64
mbert
float64
sinbert
float64
lstm_ft
float64
cnn_ft
float64
lstm_cbow
float64
cnn_cbow
float64
lstm_sl
float64
cnn_sl
float64
svm
float64
1,127,826,579,009,070,000
Singapore server එක්කට connect වෙලා YouTube ගියා පුදුම ad ගොඩක් නේ හුත්ත සිරිලංකන් ටෙලිවිසන් හොදයි
0.958382
0.962592
0.544081
0.983232
0.716309
0.724399
0.597727
0.668709
0.489622
0.552184
0.46112
1,303,906,626,555,060,000
@USER එයා එයාගෙ හුත්ත බලාගෙන පාඩුවෙ ඇති. කරදර කරන්නෙපා, පවු
0.969258
0.929525
0.537191
0.969264
0.722213
0.72542
0.392156
0.678305
0.49198
0.408784
0.243779
1,219,219,793,783,460,000
@USER @USER කියපු එකේ තේරුම කියකො සරල සිංහලෙං . . . හුත්ත
0.969573
0.904299
0.517216
0.983328
0.684283
0.726879
0.487307
0.672692
0.456316
0.485932
0.259024
713,413,707,959,767,000
අයියෝ හුත්ත මුන් හැමිනෙනවා නේ. මේකත් එදා Bangladesh මැච් එක වගේ වෙයි මේ විදිහට බිග් ශොට්ස් ගහන්න ගියොත් නම් :3... URL
0.955624
0.939852
0.538998
0.986661
0.708561
0.726109
0.503917
0.723127
0.518951
0.430649
0.293099
981,220,424,011,858,000
මේ හුත්ත ෆුල්කොත්තු තුනක් කෑව
0.963566
0.958604
0.522744
0.98291
0.721647
0.725358
0.618505
0.720125
0.628137
0.53011
0.288503
582,572,922,495,066,000
@USER @USER රත්තරන් නෙවෙයි තඹක්ක උනත් හුත්තට අනිවාර්යයෙන් දිව දාපන්- ප්ලැටිනම් උනත් හුත්ත වැරදිලාවත් උගස් තියන්න එපා. @USER @USER
0.983723
0.833747
0.521359
0.983621
0.725355
0.728576
0.691183
0.684669
0.46178
0.568747
0.278744
986,250,505,876,734,000
@USER එපාවෙලා තියෙන්නෙ බං මේ හුත්ත. ඒත් ඉතින් අමිලයගෙ වෙඩින් එක ප්ලෑන් කරන්න තියන නිසා මේක අතාරින්නත් බෑනෙ
0.971166
0.959273
0.537679
0.978879
0.655448
0.710867
0.331489
0.425536
0.564616
0.401051
0.245132
932,623,103,422,742,000
@USER හුත්ත හැරුණු අතේද? ඔලුව හැරුණු අ⁣තේද?
0.948353
0.733335
0.536611
0.973104
0.70914
0.719016
0.602939
0.535259
0.465965
0.339727
0.249771
1,112,952,842,124,310,000
@USER හික් ගෑවෙනව හුත්ත. එපා වෙනවා ඕයි. ඊට වඩා ලේසියි අනිත් පැත්තෙ එකාව ක්ෂ්ණික කෝපයෙන් මැරුව කියල මරල හිරේ ගිහිං ඇපීල් කරල එලියට එන. ටික කාලෙන් වැඩේ ඉවරයි.
0.966792
0.943046
0.53447
0.981359
0.691911
0.720296
0.567979
0.700394
0.653935
0.417111
0.589872
565,361,387,644,654,000
@USER @USER මන් මේ සිංහල හෑෂ් ටැග් වැඩ කරනවද බැලුවා :-D #හුත්ත
0.881926
0.684641
0.525106
0.369719
0.712134
0.723462
0.443266
0.648748
0.376244
0.377417
0.167323
1,144,188,886,341,550,000
@USER @USER ඔව් බන් හුත්ත
0.968413
0.964341
0.511308
0.984386
0.71745
0.729071
0.50703
0.682131
0.432662
0.442727
0.312879
877,600,392,200,167,000
මට බෑ. මට ටයිම් මැෂින් එකක් දීපං. හුත්ත. #GOT #GameOfThrones #Season7 #WinterIsHere URL
0.968288
0.940203
0.528671
0.981395
0.718189
0.729134
0.4512
0.711107
0.389583
0.407821
0.341053
1,198,613,163,421,270,000
ඒක තමයි හුත්ත ආ කමක් නෑ… @USER · 24 Nov 2019 කුනුහරප කියන එක පොරත්වයක් නොවේ
0.956008
0.96553
0.526228
0.987558
0.708078
0.725724
0.525868
0.534522
0.577907
0.471419
0.586512
1,054,672,752,873,860,000
@USER , @USER @USER හුත්ත ද රත්ත ද?
0.973868
0.93401
0.516718
0.974858
0.700701
0.724853
0.5664
0.703222
0.457095
0.448975
0.204468
1,060,222,095,706,210,000
‌පොල්පලා හුත්ත රත්තා @USER · 7 Nov 2018 මල් හුත්ත twitter.com/maalupaan/stat…
0.982178
0.939692
0.506446
0.971928
0.725678
0.7293
0.673484
0.729991
0.636429
0.543375
0.69043
857,587,543,159,837,000
ඔක්කොම ඉවසුවෑකි පත්තිනි ටෝක්ස් තමා ඉවසන්න බැරි හුත්ත
0.981899
0.969837
0.532982
0.984038
0.726881
0.728386
0.64053
0.723433
0.461842
0.432488
0.257873
1,380,712,110,443,800,000
මෙම ටුවිටය සහ එහි කෝටඩ් ටුවිටය ඇසුරින් කෝච්චි සහ හුත්ත අතර ඇති අන්තර් සබඳතාව සරලව පැහැදිලි කරන්න. ලකුනු 25 Hamza @USER · 9 Apr 2021 කොච්චි ස්ට්‍රයික්. හුත්ත twitter.com/ExcuzeMiii/sta…
0.944574
0.494425
0.529713
0.90132
0.701997
0.727271
0.526547
0.471383
0.385652
0.411465
0.381178
1,027,613,261,795,980,000
මදැයි ඉඳල ඉඳල ෆෝන් එක අන්ලොක් කරගත්ත. ඊට වඩා හොඳයි හුත්ත පඩුවෙ wifi එකකට කනෙක්ට් වෙලාම හිටියනං
0.942513
0.934784
0.534289
0.980662
0.66381
0.724222
0.450143
0.710384
0.494989
0.423664
0.306332
1,424,339,445,256,310,000
@USER @USER අනේ බන් ප්ලෙන්ටියක් බොන්න සල්ලි නෑ හුත්ත
0.971102
0.972927
0.532387
0.987362
0.715501
0.723173
0.524908
0.698653
0.512435
0.479159
0.397253
1,253,203,262,838,280,000
@USER දනා මට දුමක් දීලා කිව්වනේ ගහපන් කියලා ඇතිනේ හුත්ත
0.981649
0.965133
0.538437
0.981211
0.724498
0.723643
0.604485
0.702934
0.419678
0.373593
0.209408
913,058,502,838,497,000
එහෙම ලවු නම් එපා එපාමයි හුත්ත. තිත්ත වෙලා.
0.960592
0.969366
0.545511
0.982806
0.712306
0.725368
0.534314
0.711559
0.508533
0.476107
0.276833
965,608,197,661,843,000
දත් සෙට් එක රිදෙනවා... හුත්ත
0.976053
0.958807
0.529432
0.981865
0.723121
0.728743
0.563619
0.713223
0.399143
0.37942
0.114568
1,199,404,968,991,530,000
මොනෞත්තක්ද අයිසේ.... මං බීලත් නෑ ගංජා ගහලාත් නෑ. අමුවෙන් ඉන්නේ අද... හුත්ත පෛය
0.981373
0.972525
0.538293
0.973846
0.701968
0.720319
0.562871
0.613858
0.429751
0.385012
0.595317
998,432,877,258,752,000
සෑ වෑව් බැරිනම් මොන හුත්ත දැම්මත් අනික් උන්ට මොකද
0.942461
0.958343
0.540321
0.978592
0.723524
0.730039
0.633892
0.730948
0.535167
0.573036
0.406767
1,221,774,678,055,800,000
වැඩක් කරන්න වාඩි උනා විතරයි කරන්ට් ර්ක ගියා හුත්ත
0.977964
0.962877
0.539667
0.985423
0.709897
0.722486
0.458511
0.498208
0.450879
0.447488
0.386291
569,525,439,976,247,000
උට මන් බැන්නනේ :'( මට අවුල් හුත්ත උබට වඩා.. >:( :'(
0.971903
0.953241
0.512845
0.988246
0.724985
0.728326
0.419884
0.55275
0.3357
0.373697
0.435011
1,229,713,812,510,690,000
@USER කාගෙ හුත්ත ගැනද කීවෙ බන්
0.962328
0.960403
0.515857
0.985156
0.710446
0.719658
0.470802
0.716858
0.483278
0.386021
0.376183
1,085,905,866,107,190,000
දැන් නම් මේ හුත්ත එපා වෙලා තියෙන්නේ
0.954061
0.953588
0.531789
0.984656
0.724238
0.727443
0.548513
0.726987
0.455305
0.533727
0.284237
1,380,376,746,747,250,000
Blackwhite හුත්ත ආයෙ ඇවිත් . ඇයි බං හැමදාම එකම
0.951697
0.924051
0.536805
0.983397
0.714228
0.728468
0.40596
0.57081
0.539237
0.438404
0.145424
896,424,210,611,946,000
@USER හුත්ත කිව්වෙ සමනලයට වෙන්න ඇති අලුත් වචනයක්
0.896663
0.911451
0.531456
0.98357
0.710327
0.720518
0.599916
0.705036
0.490337
0.416329
0.358951
1,445,064,947,453,500,000
අපරාදෙ හුත්ත පැකේජ් එක ඉවරවෙලා කියල හිතං පැකේජ් එකත් දැම්ම. තව සති දෙකක් ඒකෙත් ඉතුරුව තිබිලනෙ #Wtf
0.927077
0.88168
0.546075
0.984355
0.704153
0.724649
0.430921
0.671468
0.625733
0.445302
0.217937
897,351,543,497,912,000
මොනා උනත් අපේ තත්වෙ තියාගන්න ඕනෙනෙ හුත්ත URL
0.959371
0.899445
0.524982
0.987046
0.689499
0.728763
0.488024
0.728199
0.450932
0.467874
0.35305
662,142,416,846,086,000
@USER @USER සිරා සීන් එක බ්‍රෝ! ගු හුත්තල හින්ද තමයි දැන් ඕකත් muscial ෂෝ ගානට වැටෙන්නේ! හුත්ත!
0.986578
0.951187
0.513254
0.982741
0.725165
0.729287
0.422328
0.711507
0.448442
0.457808
0.277211
1,206,075,336,108,080,000
@USER @USER තිරිසන් වැඩ හුත්ත
0.983072
0.966004
0.511308
0.988421
0.724244
0.729643
0.691577
0.730405
0.424693
0.413827
0.265256
1,341,444,404,817,480,000
අනේ හුත්ත තොපෙ අයිෆෝන් රෝල් කරල ගහගනියව්. මල කන්දොස්කිරියාවක් නෙහ් ඔයි
0.985055
0.974805
0.546821
0.985796
0.726318
0.730555
0.700732
0.728218
0.68467
0.504425
0.532693
1,271,102,195,539,130,000
මං මේකේ කුනුහරප කියනකොට බනින්න උන් දාහක් ඉන්නවා. ඒත් දැන් මං නොකියා ඉන්න නිසා හොද ලමයා කියන්න එක බල්ලෙක් නෑ හුත්ත
0.981405
0.96521
0.542321
0.976799
0.710239
0.728658
0.460246
0.725593
0.544914
0.551644
0.820913
1,061,133,737,705,000,000
මිදි කොහෙද තිත්ත නරි දන්න හුත්ත
0.984217
0.969013
0.52577
0.976906
0.726755
0.728312
0.618102
0.725028
0.389143
0.467393
0.412907
696,238,450,882,826,000
මොන හුත්ත උනත් අන්තිමට වෙන්නෙ එක දෙයයි. -_-
0.932409
0.958975
0.548004
0.986401
0.716152
0.724389
0.423778
0.720262
0.575604
0.420533
0.339945
1,132,621,432,821,670,000
Lunch එකට පලතුරු කෑවා තිබහට කොමඩු කෑවා හුත්ත ඇගට පන නෑ burger එකක්වත් කන්න ඕනේ
0.936562
0.887561
0.536748
0.962532
0.464829
0.714634
0.395577
0.627945
0.422676
0.393118
0.272303
1,478,303,486,475,330,000
මේ ටික්ටොක් ඇප් එක තහනම් කරපන්කො හුත්ත
0.984415
0.967487
0.531789
0.984754
0.726582
0.729515
0.498956
0.729094
0.49618
0.416401
0.282147
932,617,750,677,246,000
යනවා හුත්ත ඔලුව හැරුන අතේ....මොන හුත්තක්ද මේ...
0.979792
0.955418
0.533569
0.984756
0.72808
0.72904
0.677938
0.723507
0.479499
0.369234
0.66885
1,054,537,998,862,310,000
කලිං ට්විට් එකක කියපු වාහන සංවාදෙ අතරෙදිම කියවුන ප්‍රකාශයක් . . . ලංකාවේ වාහන ගෙන්නන ක්‍රමේ සුමනක්කගේ හුත්ත වගේ මල්ලි, එකලාසයක් නෑ
0.967322
0.911223
0.542969
0.973
0.583928
0.720649
0.598978
0.725646
0.418986
0.461781
0.357171
1,332,954,790,028,980,000
අනේ හුත්ත මේකේ ටයිපින් කන්ටෙන්ට් මදි. කවද්ද මේක වැඩි කරන්නේ @USER
0.950937
0.964931
0.537223
0.982862
0.710983
0.727395
0.549653
0.716775
0.505134
0.428699
0.454384
1,104,227,768,320,100,000
දැන්නම් පොලීසිය දැක්කොත් හෙඩ් එක ගහලා කියනවාමයි ඇයි හුත්ත වාහනේ ගන්නකොට බදුත් ගහලා තව පදින කොට දඩත් ගහන්න
0.84134
0.942701
0.540947
0.970065
0.516453
0.726199
0.549961
0.726919
0.424622
0.420499
0.370207
751,808,630,073,102,000
මොන හුත්ත දැම්මාත් කමක් නෑ 10.30 ට Prometheus බලන්න දෙන්න ඕනේ -_-
0.959842
0.964595
0.531758
0.983613
0.709317
0.726731
0.55823
0.666305
0.482612
0.36754
0.251943
1,475,871,886,357,660,000
කන් දෙක රිදෙනව හුත්ත මේ මාස්ක් එකට
0.96069
0.960262
0.531789
0.984904
0.725972
0.727352
0.547892
0.706279
0.385504
0.408663
0.393532
1,121,471,986,813,650,000
ජමියතුල් උලමා එකෙන් ත්‍රස්තවාදීන්ගේ මල සිරුරු බාරගන්නෙ නැ කිව්වම ඒකට තියන කමෙන්ට් එකක් "භාරගන්න මිනී නෑ අරඅරූගේ ජුන්ඩා තැපැල් කරලා නාසායවපං චන්ද්‍රිකාවක්වත් යන වෙලාවක අරින්ඩ කියලා ස්වර්ගෙට ඒ හුත්ත ඇති කන්‍යාවෝ 72 දිහා බලන් ජුන්ඩා නැතුව පලංචියට අහුවෙලා ජුන්ඩා කඩාගත්ත රිලවා වගේ
0.983868
0.79493
0.537537
0.413167
0.575169
0.7223
0.581443
0.667534
0.577355
0.414767
0.60559
1,257,556,119,532,300,000
අනේ හුත්ත මේක හරි අසාධාරනයි අර ශර්ලොක් හෝම්ස් මාව බොක් කරලනේ මම ඌට මොනවත් කියන්නෙත් නැතිව
0.925318
0.94293
0.545591
0.971371
0.70741
0.725427
0.455899
0.609273
0.577684
0.420651
0.777529
830,970,900,396,667,000
අපි SM වල කුණුහර්ප දාපු හැටි. - @#$% - හු#$% - හු*ත - හුත්ත
0.985822
0.938579
0.532741
0.981114
0.723815
0.728576
0.617793
0.730861
0.339531
0.347259
0.509631
712,698,930,186,883,000
අනේ හුත්ත... අතේ තිබ්බ මැච් එකනෙ..
0.973995
0.965258
0.545447
0.985295
0.707776
0.720633
0.595488
0.616783
0.516523
0.430721
0.260078
951,674,038,597,775,000
මොනා හරි කරන වෙනවා හැම එකාම ඔලුවට අත හෝදන්න එනවානෙ හුත්ත
0.920861
0.942451
0.551008
0.982033
0.721383
0.728498
0.512047
0.706695
0.371716
0.430089
0.328919
946,657,736,707,670,000
මාත් Repeat ටෞකන්ඩ කරන්නෙ නෑ මේ පාර...හුත්ත
0.97755
0.582081
0.536762
0.979593
0.284405
0.282006
0.307959
0.424232
0.378689
0.296258
0.613475
856,570,482,224,583,000
@USER තව මාසෙකින් ඕයි. එතකොට හුත්ත මෙහෙ එලියට බහින්ඩ බෑ වහිනව
0.933169
0.927193
0.537915
0.985148
0.71858
0.728181
0.595721
0.637426
0.546412
0.454944
0.320828
885,917,653,910,438,000
@USER අපරාදෙ හුත්ත ඩේට ටික
0.973371
0.962354
0.512795
0.988594
0.714989
0.724358
0.409053
0.715182
0.483357
0.368122
0.198423
1,253,250,847,812,830,000
වලි නම් වලි දාගනිල්ලා වලියට ගිහින් ආතල් විහිලු කියලා කියන්න එපා කන්ෆියුස් වෙනවනෙ හුත්ත
0.947115
0.962139
0.548562
0.97792
0.614888
0.721419
0.571609
0.716946
0.45549
0.441002
0.47234
767,514,529,408,225,000
මම යනවා නිදාගන්න හුත්ත දැන් ඕක බලලා වැඩක් නෑ 󾍀 Sasha පැරදුනා 󾌹 සිකේ
0.963623
0.958713
0.544567
0.986657
0.664957
0.717205
0.350715
0.687907
0.413802
0.381499
0.298564
1,018,800,235,416,560,000
@USER මේ රිප්ලයි එක notification වල නෑ හුත්ත
0.970056
0.966167
0.537058
0.984161
0.717195
0.726195
0.314266
0.622721
0.445017
0.353841
0.33185
1,199,668,509,187,260,000
කෝච්චි 02 හැප්පුනොත් අනතුරක් වන අතර බිංගෙයක ලස්සන දකින්නේ තවත් බිංගයක් නොවන බැව් වටහා ගත යුතුය. කොටින්ම කිව්වොත්, ලිංග දෙකකින් ජීවීන් වර්ග දෙකක් ඉන්නේ අහවලෞත්තකට නොවන අතර පෛය සහ හුත්ත කියන 02 ජීවින්ට දීලා තියෙන්නේ බර දා හිකීමට මිස එකම දෙකක් අතුල්ල අතුල්ල ඉන්න නොවෙයි. ස්තූතියි.
0.923518
0.928735
0.537123
0.91047
0.358285
0.721136
0.466637
0.295331
0.470686
0.448045
0.224893
648,535,093,967,884,000
හු*ත හුත්ත මැදට තරු එබුවම සභ්‍ය වෙන්නේ කොහොමද බං -Saw on Fb-
0.974562
0.95429
0.52227
0.983557
0.708169
0.720868
0.592534
0.649831
0.552465
0.44315
0.322398
1,109,640,508,307,960,000
@USER @USER ඔය හුත්ත කියල කියල යාළුවෙකුත් තරහ උනා. මේකටත් බෝ කරපංකෝ.
0.944405
0.910606
0.522348
0.986593
0.679295
0.72279
0.456517
0.432863
0.493812
0.40794
0.601533
1,294,877,407,405,440,000
උබ මෙකෙ මෙහෙම නහිනවා කියලා දන්නවා නම් ගෝඨට චන්දෙ දෙන්නෙත් නෑ පාර්ලිමෙන්තු මැතිවරනෙදි පොහොට්ටුවට කතිරෙ ගහන්නෙ නෑ ඔයි හුත්ත
0.980001
0.96914
0.545778
0.98872
0.719463
0.727635
0.61629
0.721115
0.461381
0.403773
0.376943
1,327,640,779,624,300,000
ebay account එකක් හදා ගත්තු ඔන්ටප්‍රනර්ස් ලා දැකලම එපා වෙලා හුත්ත
0.980006
0.8062
0.525855
0.984389
0.717153
0.724552
0.511612
0.691389
0.487646
0.417621
0.264322
1,133,659,727,362,060,000
දුෂණ වංචා ජරාව කොමිස් නෙමෙයි බන් මේ තුවාලෙම හාර හාර වනයක් කරන හැටි ලියපන් එහෙම නැත්නම් මේ හුත්ත ආයේමත් කනත්තක් වෙයි ඔවි, අපි ඒ ගැනත් ලියමු සහොදරයා #සහොදරයා
0.960522
0.950588
0.544337
0.982638
0.451597
0.716582
0.687969
0.682403
0.697696
0.476891
0.5
1,162,029,665,357,320,000
"මිදි කොහේද තිත්ත - නරි දන්න හුත්ත" - බුකියෙන්
0.984599
0.961065
0.54743
0.978896
0.725477
0.726951
0.65245
0.723611
0.402175
0.514648
0.412907
1,250,880,525,482,570,000
අද මේකෙ ඉන්නවා එකෙක් ලෛ එන්නෙ නෑ.අපිට පොඩ්ඩක් නින්ද යනවා නේද මුන් ලෛ ඇවිල්ලා සහනාධාරත් බෙදලා ගෙවලුත් හදලා දෙනවා හුත්ත
0.931227
0.954703
0.540146
0.983075
0.692827
0.71972
0.546993
0.729112
0.42111
0.492202
0.246962
1,060,222,711,832,600,000
පැණි කොමඩු හුත්ත සුදු මාමේ දුව දෙනවද @USER · 7 Nov 2018 ‌පොල්පලා හුත්ත twitter.com/Rathgamaya/sta…
0.984196
0.733932
0.518029
0.97338
0.716795
0.7268
0.598633
0.726649
0.690636
0.523079
0.646638
910,189,633,115,374,000
@USER එපා කියපං බං. උන්ට කියපං ACBT හරි KDU හරි කරන්ඩ කියල. තේරුමක් නෑ මෙ හුත්ත.
0.973022
0.969215
0.520899
0.984171
0.691173
0.724937
0.514499
0.729369
0.571749
0.423073
0.267551
1,052,419,621,310,140,000
@USER @USER ඔය ඉතින් . . . ආයෙත් හුත්ත ලු. මේ හුත්ත කරන්න බෑ මේ හුත්තලත් එක්ක. හුත්ත !
0.979529
0.939815
0.522402
0.985825
0.729534
0.729824
0.700746
0.72686
0.645739
0.543846
0.308622
673,535,199,389,483,000
හුත්ත.. කැරිය වගේ ලොකු ලයින් දුන්නට වැටෙන් නැතුව බයික් එකක් වත් එලවගන්න බැරිද බොට. :3 :3
0.974408
0.921858
0.534399
0.986874
0.715823
0.653225
0.558944
0.563378
0.597122
0.419339
0.308314
649,023,246,482,829,000
හුත්ත #Instagram උබ මගේ දව්ස කෑවා.
0.979031
0.954285
0.511241
0.984417
0.723572
0.706384
0.602872
0.656073
0.451991
0.361126
0.257974
1,182,301,582,441,900,000
මිනී මරාගන්න තමයි ලෑස්ති වෙන්නේ හුත්ත
0.982158
0.951403
0.52577
0.983994
0.7237
0.723293
0.566013
0.470761
0.456761
0.487979
0.329643
1,015,219,380,701,910,000
ඉස්සර කොහොම කුණුහරුප කියන් හිටපු මමද බං හුත්ත දැන් මේ හොඳ වෙලා තියෙන්නේ ඒවා මතක උන් තාම ඉන්නවද
0.920995
0.961055
0.540947
0.98332
0.670818
0.729051
0.518249
0.727171
0.678578
0.612134
0.405316
732,818,385,650,786,000
අපි ෆෝන් චාර්ජ් නොකරන දවසටම තමයි කරන්ට් එකත් යන්නේ හුත්ත -_- 6% :3
0.920028
0.958666
0.527471
0.986943
0.69699
0.708952
0.397292
0.675447
0.423773
0.483033
0.356443
1,240,536,417,908,010,000
වැඩක් කරගන්නත් නෑ හුත්ත සිග්නල් නෑ හරියට @USER @USER
0.933518
0.942712
0.532165
0.987382
0.717846
0.726514
0.349994
0.620254
0.36907
0.351839
0.34479
1,054,721,528,795,820,000
@USER @USER අනේ හුත්ත මම විතරක් කියපු දෙයක් නෙවෙයි ගයාන් අයියත් කිව්වා බොංගො එකේ සීන් එක
0.95341
0.959241
0.531213
0.979924
0.700119
0.725429
0.661679
0.691713
0.580671
0.463696
0.349569
921,775,456,361,308,000
කොමත් ශරීරගත වෙන ඩෝසේජ් එක එක්ක සුවර් නෑ අනික කොමද බං දන්නෙ මොන හුත්ත වෙල මැරෙයිද කියල. අනිත්‍යයි
0.931034
0.958826
0.544906
0.966232
0.663864
0.7231
0.579073
0.708806
0.418157
0.592531
0.46224
1,042,116,512,008,390,000
@USER හුත්ත බලන්න ආසයි බ්‍රො
0.966395
0.954404
0.512795
0.983386
0.702128
0.712749
0.556265
0.568645
0.541402
0.408908
0.205289
1,052,421,125,668,580,000
@USER මොන හුත්තක්ද අයියේ...අනික දැන් ඔය හුත්ත එච්චරම හැංගිලා තියෙන දෙයක් නෙවෙයිනේ ඕන හුත්තක් වෙන්න කියලා හැම හුත්තිම එලියේ දාන් යනෝනේ හුත්ත
0.983889
0.939792
0.537989
0.987659
0.729461
0.729748
0.69196
0.717214
0.648427
0.461505
0.820062
852,557,444,584,808,000
හුත්ත මරන තර්ජන එනවනෙ. බෑනෙ කරන්ඩ
0.959358
0.930671
0.526046
0.990027
0.722401
0.685703
0.535348
0.667431
0.497517
0.430855
0.475158
1,211,733,379,705,650,000
පාන්දර 1යි දැන් වෙලාව..සැප නින්දක හිටියෙ.හිටියෙ. කව්ද පොරක් කෝල් කරලා සොරි වැරදීමක් කියලා තිබ්බා හුත්ත.
0.951599
0.913237
0.535386
0.957735
0.697271
0.720138
0.383029
0.718668
0.429305
0.494978
0.437139
1,071,395,840,395,280,000
අයියෝ ෆෙමිනිස් නංගිලා ..මෙන්න ඔයාලගෙ ලොකු අක්කා කටේ අතුල්ලලා දැන්නම් එන්න එපා ඔන්නේ ඒක එයාගෙ කැමැත්ත හුත්ත කියාගෙන ... හොද අත්තක් අල්ලනකම් විතරයි ෆෙමිනිස් කම් Deepika Narayan Bhardwaj @USER · 7 Dec 2018 If adding husbands name, wearing sindoor, mangal sutra, chuda are things that feminists actually loveadorn at the first go as identity of a married woman, why brainwash the entire young generation of girls against it? Why brand them as tools of oppression? Amazed !!!! Show this thread
0.967725
0.922117
0.504634
0.970626
0.404622
0.713409
0.392257
0.681513
0.578408
0.510546
0.539667
1,370,317,495,924,530,000
මේ හුත්ත යන්නෑ මචන් කියලා අයියා කෙනෙක් බැහැලා ගියා...
0.972016
0.959566
0.54842
0.985884
0.719871
0.72258
0.446231
0.48608
0.477561
0.419022
0.311946
1,212,010,819,103,460,000
අනේ හුත්ත මටත් මගේ හිතේ තියන දේ කියන්න ඕනේ
0.94563
0.950579
0.540321
0.982641
0.714011
0.724995
0.595798
0.584469
0.495145
0.429509
0.425966
851,106,454,148,919,000
@USER @USER අනේ හුත්ත මං මේ වලියක් කියල කන එකත් නවත්තල ආව
0.936244
0.968799
0.530489
0.984738
0.718363
0.729466
0.466015
0.723395
0.615376
0.569339
0.46246
1,126,482,734,464,030,000
අර කැරි BlockEvent හුත්ත නිසයි මේ ඔක්කොම. #IO19LK
0.974949
0.961483
0.535475
0.98471
0.729139
0.729688
0.704265
0.731036
0.623484
0.533386
0.480736
867,024,290,016,666,000
@USER @USER ගෝතයෝ, ආසවෙන් බලපු පේජ් එකක්. දැන් ඒක.කරන ගොං හුත්ත ඒක කනව.
0.982301
0.96469
0.522605
0.986883
0.715669
0.728256
0.650236
0.729055
0.433642
0.444405
0.3213
1,313,189,717,345,700,000
@USER හුත්ත කේන්දරේ තිබ්බොත් ලොකු අවුලක් නේද ?
0.935987
0.902199
0.531267
0.98658
0.711762
0.722071
0.396246
0.690651
0.522054
0.367565
0.271662
1,139,858,202,231,660,000
පව් කියලා මැසේජ් එකක් දැම්මහම එකත් ලොකු සීන්. ඒමඋං නං ඕනෙම නෑ හුත්ත
0.971855
0.947895
0.54806
0.985134
0.711727
0.72242
0.340908
0.599725
0.463249
0.359369
0.289405
886,158,234,838,458,000
මුන්ට හුත්ත කුණුහරුප කියන්නත් ඕන, හැබැයි "ගුරු සිත නොරිදවා.. වේලාව නොවරදවා.. බැති පෙම් උපදවා" ඒ හුත්ත කරන්නත් ඕන.... URL
0.975894
0.910807
0.533107
0.985054
0.726661
0.729002
0.691084
0.38815
0.678601
0.528189
0.230804
1,173,584,941,487,020,000
@USER @USER ආච්චිගේ හුත්ත.කටවහප් පරයා
0.984832
0.971308
0.526604
0.984742
0.300874
0.309357
0.460244
0.358785
0.40645
0.37444
0.888749
1,361,268,842,073,260,000
මිදි කොහෙද තිත්ත නරි දන්න හුත්ත Nimila Hiranya @USER · 14 Feb 2021 Only two modes.
0.982439
0.965226
0.503525
0.972163
0.721269
0.726865
0.633671
0.730132
0.415702
0.475975
0.450621
1,473,642,348,382,870,000
@USER ඊයා හුත්ත කුණුහර්ප කියනව මෙයා
0.978224
0.973779
0.515857
0.983199
0.715401
0.726055
0.601006
0.715472
0.431375
0.359659
0.298484
1,152,249,875,184,030,000
@USER මේ හුත්ත හරියන්නෑ ජීවිතේ මුල ඉඳල පටන් ගන්න ඕන
0.927089
0.953235
0.538437
0.983941
0.706105
0.722175
0.388885
0.348186
0.579263
0.455819
0.241603
1,092,732,154,373,530,000
ගෑනු ඔක්කොම එකයි ... එපා වෙනවා හුත්ත
0.977704
0.97181
0.535601
0.98526
0.725994
0.728291
0.484338
0.665439
0.489308
0.480717
0.39573
845,646,278,318,948,000
මම කුනුහර්ප කියන එක . කෝමහරි නවත්තනවා.... . . . . . . . . . . . . . . . . . . . . . . හුත්ත
0.983433
0.881092
0.538066
0.961264
0.719849
0.728004
0.533868
0.729386
0.409872
0.397733
0.378206
1,295,212,958,608,500,000
@USER හුත්ත හුත්ත කියන ටික විතරයි තේරුණේ.
0.982579
0.928215
0.532873
0.982462
0.729449
0.729578
0.694006
0.726396
0.62756
0.462781
0.340016
703,067,870,717,211,000
මේ දැන් කරන්ට් එක ගියා -_- හුත්ත රෑට නිදාගන්න කලින් ෆෝන් එක Charge වෙන්න දාපු දවසට මොන හුත්තක් වත් වෙන්නේ නෑ..... URL
0.951478
0.945034
0.533925
0.984045
0.72633
0.725906
0.379491
0.692084
0.652005
0.524591
0.648039
1,113,208,019,913,920,000
යන යන ඇදුම් කඩ වල කෙල්ලො ඉන්නෙ ඒයි ජොකෙක් ගන්න ගියත් මොකක්ද සයිස් එක කියල අහනවා හුත්ත හැදියාවක් නැති හැටි
0.959514
0.955648
0.543647
0.975197
0.723963
0.726558
0.640095
0.609492
0.491352
0.438859
0.475939
1,173,315,146,065,700,000
මොකාද් හුත්ත නිසල් මඩුශාන් කියන්නෙ
0.983351
0.941077
0.522744
0.982493
0.713173
0.725942
0.60592
0.729338
0.432304
0.372923
0.261412
658,694,364,944,687,000
අයි ගනු සරියෙ හුත්ත ගාව තද කරන්නෙ ????
0.980813
0.96654
0.548719
0.981984
0.718585
0.727799
0.570561
0.565216
0.489312
0.53048
0.262277

SOLD - A Benchmark for Sinhala Offensive Language Identification

In this repository, we introduce the {S}inhala {O}ffensive {L}anguage {D}ataset (SOLD) and present multiple experiments on this dataset. SOLD is a manually annotated dataset containing 10,000 posts from Twitter annotated as offensive and not offensive at both sentence-level and token-level. SOLD is the largest offensive language dataset compiled for Sinhala. We also introduce SemiSOLD, a larger dataset containing more than 145,000 Sinhala tweets, annotated following a semi-supervised approach.

:warning: This repository contains texts that may be offensive and harmful.

Annotation

We use an annotation scheme split into two levels deciding (a) Offensiveness of a tweet (sentence-level) and (b) Tokens that contribute to the offence at sentence-level (token-level).

Sentence-level

Our sentence-level offensive language detection follows level A in OLID (Zampieri et al., 2019). We asked annotators to discriminate between the following types of tweets:

  • Offensive (OFF): Posts containing any form of non-acceptable language (profanity) or a targeted offence, which can be veiled or direct. This includes insults, threats, and posts containing profane language or swear words.
  • Not Offensive (NOT): Posts that do not contain offense or profanity.

Each tweet was annotated with one of the above labels, which we used as the labels in sentence-level offensive language identification.

Token-level

To provide a human explanation of labelling, we collect rationales for the offensive language. Following HateXplain (Mathew et al., 2021), we define a rationale as a specific text segment that justifies the human annotator’s decision of the sentence-level labels. Therefore, We ask the annotators to highlight particular tokens in a tweet that supports their judgement about the sentence-level label (offensive, not offensive). Specifically, if a tweet is offensive, we guide the annotators to highlight tokens from the text that supports the judgement while including non-verbal expressions such as emojis and morphemes that are used to convey the intention as well. We use this as token-level offensive labels in SOLD.

Alt text

Data

SOLD is released in HuggingFace. It can be loaded in to pandas dataframes using the following code.

from datasets import Dataset
from datasets import load_dataset

sold_train = Dataset.to_pandas(load_dataset('sinhala-nlp/SOLD', split='train'))
sold_test = Dataset.to_pandas(load_dataset('sinhala-nlp/SOLD', split='test'))

The dataset contains of the following columns.

  • post_id - Twitter ID
  • text - Post text
  • tokens - Tokenised text. Each token is seperated by a space.
  • rationals - Offensive tokens. If a token is offensive it is shown as 1 and 0 otherwise.
  • label - Sentence-level label, offensive or not-offensive.

Alt text

SemiSOLD is also released HuggingFace and can be loaded to a pandas dataframe using the following code.

from datasets import Dataset
from datasets import load_dataset

semi_sold = Dataset.to_pandas(load_dataset('sinhala-nlp/SemiSOLD', split='train'))

The dataset contains following columns

  • post_id - Twitter ID
  • text - Post text

Furthermore it contains predicted offensiveness scores from nine classifiers trained on SOLD train; xlmr, xlmt, mbert, sinbert, lstm_ft, cnn_ft, lstm_cbow, cnn_cbow, lstm_sl, cnn_sl and svm

Experiments

Clone the repository and install the libraries using the following command (preferably inside a conda environment)

pip install -r requirements.txt

Sentence-level

Sentence-level transformer based experiments can be executed using the following command.

python -m experiments.sentence_level.sinhala_deepoffense

The command takes the following arguments;

--model_type : Type of the transformer model (bert, xlmroberta, roberta etc ).
--model_name : The exact architecture and trained weights to use. This may be a Hugging Face Transformers compatible pre-trained model, a community model, or the path to a directory containing model files.
--transfer : Whether to perform transfer learning or not (true or false).
--transfer_language : The initial language if transfer learning is performed (hi, en or si).
    * hi - Perform transfer learning from HASOC 2019 Hindi dataset (Modha et al., 2019).
    * en - Perform transfer learning from Offenseval English dataset (Zampieri et al., 2019).
    * si - Perform transfer learning from CCMS Sinhala dataset (Rathnayake et al., 2021).
--augment : Perform semi supervised data augmentation.
--std : Standard deviation of the models to cut down data augmentation.
--augment_type: The type of the data augmentation.
    * off - Augment only the offensive instances.
    * normal - Augment both offensive and non-offensive instances.

Sentence-level CNN and LSTM based experiments can be executed using the following command.

python -m experiments.sentence_level.sinhala_offensive_nn

The command takes the following arguments;

--model_type : Type of the architecture (cnn2D, lstm).
--model_name : The exact word embeddings to use. This may be a gensim model, or the path to a word embeddinng files.
--augment : Perform semi supervised data augmentation.
--std : Standard deviation of the models to cut down data augmentation.
--augment_type: The type of the data augmentation.
    * off - Augment only the offensive instances.
    * normal - Augment both offensive and non-offensive instances.

Token-level

Token-level transformer based experiments can be executed using the following command.

python -m experiments.sentence_level.sinhala_mudes

The command takes the following arguments;

--model_type : Type of the transformer model (bert, xlmroberta, roberta etc ).
--model_name : The exact architecture and trained weights to use. This may be a Hugging Face Transformers compatible pre-trained model, a community model, or the path to a directory containing model files.
--transfer : Whether to perform transfer learning or not (true or false).
--transfer_language : The initial language if transfer learning is performed (hatex or tsd).
    * hatex - Perform transfer learning from HateXplain dataset (Mathew et al., 2021).
    * tsd - Perform transfer learning from TSD dataset (Pavlopoulos  et al., 2021).

Token-level LIME experiments can be executed using the following command.

python -m experiments.sentence_level.sinhala_lime

The command takes the following arguments;

--model_type : Type of the transformer model (bert, xlmroberta, roberta etc ).
--model_name : The exact architecture and trained weights to use. This may be a Hugging Face Transformers compatible pre-trained model, a community model, or the path to a directory containing model files.

Acknowledgments

We want to acknowledge Janitha Hapuarachchi, Sachith Suraweera, Chandika Udaya Kumara and Ridmi Randima, the team of volunteer annotators that provided their free time and efforts to help us produce SOLD.

Citation

If you are using the dataset or the models please cite the following paper

@article{ranasinghe2022sold,
  title={SOLD: Sinhala Offensive Language Dataset},
  author={Ranasinghe, Tharindu and Anuradha, Isuri and Premasiri, Damith and Silva, Kanishka and Hettiarachchi, Hansi and Uyangodage, Lasitha and Zampieri, Marcos},
  journal={arXiv preprint arXiv:2212.00851},
  year={2022}
}
Downloads last month
2
Edit dataset card