Sentiment_Analysis_3W0623 / reports /evaluation_comparison.json
Lavender825's picture
Deploy 10W0715 dual-interface sentiment app
422d4ca
Raw
History Blame Contribute Delete
18.6 kB
{
"overall_3class_comparison": {
"Baseline_1_TFIDF_LogReg": {
"macro_f1": 0.6227417071292987,
"accuracy": 0.8090666666666667
},
"Baseline_2_BERT_overall_3class": {
"macro_f1": 0.7284109895875363,
"accuracy": 0.8962
},
"Baseline_3_BERT_ACSA_no_meta__aggregated_to_overall": {
"macro_f1": 0.503449963221516,
"accuracy": 0.6606
},
"Proposed_BERT_Meta_Fusion__overall_head": {
"macro_f1": 0.7176867208554908,
"accuracy": 0.8942666666666667
},
"Proposed_BERT_Meta_Fusion__aggregated_to_overall_(reference)": {
"macro_f1": 0.5224811910366102,
"accuracy": 0.6916666666666667
}
},
"proposed_per_aspect": {
"per_aspect": {
"SIZE": {
"macro_f1": 0.8688511032834313,
"accuracy": 0.876,
"confusion_matrix": [
[
3919,
392,
177
],
[
650,
6782,
397
],
[
138,
106,
2439
]
],
"report": {
"Not_Mentioned": {
"precision": 0.8325897599320161,
"recall": 0.8732174688057041,
"f1-score": 0.8524197933659597,
"support": 4488.0
},
"Positive": {
"precision": 0.9315934065934066,
"recall": 0.8662664452675949,
"f1-score": 0.8977430670461315,
"support": 7829.0
},
"Negative": {
"precision": 0.8094922004646532,
"recall": 0.9090570257174805,
"f1-score": 0.8563904494382022,
"support": 2683.0
},
"accuracy": 0.876,
"macro avg": {
"precision": 0.857891788996692,
"recall": 0.8828469799302598,
"f1-score": 0.8688511032834313,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8801316797760889,
"recall": 0.876,
"f1-score": 0.8767857386915525,
"support": 15000.0
}
}
},
"MATERIAL": {
"macro_f1": 0.8351109775214702,
"accuracy": 0.8648666666666667,
"confusion_matrix": [
[
3525,
486,
247
],
[
577,
7811,
446
],
[
176,
95,
1637
]
],
"report": {
"Not_Mentioned": {
"precision": 0.8239831697054698,
"recall": 0.8278534523250353,
"f1-score": 0.8259137769447048,
"support": 4258.0
},
"Positive": {
"precision": 0.930767397521449,
"recall": 0.8841974190627122,
"f1-score": 0.9068849413677,
"support": 8834.0
},
"Negative": {
"precision": 0.7025751072961374,
"recall": 0.8579664570230608,
"f1-score": 0.7725342142520056,
"support": 1908.0
},
"accuracy": 0.8648666666666667,
"macro avg": {
"precision": 0.819108558174352,
"recall": 0.8566724428036028,
"f1-score": 0.8351109775214702,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8714288554020935,
"recall": 0.8648666666666667,
"f1-score": 0.8668105143377094,
"support": 15000.0
}
}
},
"QUALITY": {
"macro_f1": 0.7742721084444115,
"accuracy": 0.8063333333333333,
"confusion_matrix": [
[
6322,
1106,
462
],
[
958,
4841,
190
],
[
141,
48,
932
]
],
"report": {
"Not_Mentioned": {
"precision": 0.85190675111171,
"recall": 0.8012674271229404,
"f1-score": 0.8258115080660963,
"support": 7890.0
},
"Positive": {
"precision": 0.8075062552126773,
"recall": 0.8083152446151277,
"f1-score": 0.8079105473965287,
"support": 5989.0
},
"Negative": {
"precision": 0.5883838383838383,
"recall": 0.8314005352363961,
"f1-score": 0.68909426987061,
"support": 1121.0
},
"accuracy": 0.8063333333333333,
"macro avg": {
"precision": 0.7492656149027419,
"recall": 0.8136610689914882,
"f1-score": 0.7742721084444115,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8144851674378932,
"recall": 0.8063333333333333,
"f1-score": 0.8084469162349509,
"support": 15000.0
}
}
},
"APPEARANCE": {
"macro_f1": 0.7866997719239698,
"accuracy": 0.8634666666666667,
"confusion_matrix": [
[
7598,
606,
366
],
[
783,
4841,
131
],
[
118,
44,
513
]
],
"report": {
"Not_Mentioned": {
"precision": 0.893987527944464,
"recall": 0.886581096849475,
"f1-score": 0.8902689085476595,
"support": 8570.0
},
"Positive": {
"precision": 0.8816244764159534,
"recall": 0.8411815812337098,
"f1-score": 0.8609283300729148,
"support": 5755.0
},
"Negative": {
"precision": 0.5079207920792079,
"recall": 0.76,
"f1-score": 0.6089020771513353,
"support": 675.0
},
"accuracy": 0.8634666666666667,
"macro avg": {
"precision": 0.7611775988132085,
"recall": 0.8292542260277282,
"f1-score": 0.7866997719239698,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8718712340607555,
"recall": 0.8634666666666667,
"f1-score": 0.8663503991933479,
"support": 15000.0
}
}
},
"STYLE": {
"macro_f1": 0.7622600117091981,
"accuracy": 0.8287333333333333,
"confusion_matrix": [
[
6595,
823,
438
],
[
911,
5219,
235
],
[
115,
47,
617
]
],
"report": {
"Not_Mentioned": {
"precision": 0.8653719984254035,
"recall": 0.8394857433808554,
"f1-score": 0.8522323447696583,
"support": 7856.0
},
"Positive": {
"precision": 0.8571193956314666,
"recall": 0.8199528672427338,
"f1-score": 0.8381242974144854,
"support": 6365.0
},
"Negative": {
"precision": 0.4782945736434108,
"recall": 0.79204107830552,
"f1-score": 0.5964233929434509,
"support": 779.0
},
"accuracy": 0.8287333333333333,
"macro avg": {
"precision": 0.7335953225667603,
"recall": 0.8171598963097031,
"f1-score": 0.7622600117091981,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8417679230461648,
"recall": 0.8287333333333333,
"f1-score": 0.8329608184437722,
"support": 15000.0
}
}
},
"VALUE": {
"macro_f1": 0.7179350422128157,
"accuracy": 0.8483333333333334,
"confusion_matrix": [
[
10218,
1191,
372
],
[
511,
2162,
95
],
[
72,
34,
345
]
],
"report": {
"Not_Mentioned": {
"precision": 0.9460235163410795,
"recall": 0.8673287496816908,
"f1-score": 0.9049685590293154,
"support": 11781.0
},
"Positive": {
"precision": 0.6383229997047535,
"recall": 0.7810693641618497,
"f1-score": 0.7025182778229082,
"support": 2768.0
},
"Negative": {
"precision": 0.4248768472906404,
"recall": 0.7649667405764967,
"f1-score": 0.5463182897862233,
"support": 451.0
},
"accuracy": 0.8483333333333334,
"macro avg": {
"precision": 0.6697411211121578,
"recall": 0.8044549514733458,
"f1-score": 0.7179350422128157,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8735733711550063,
"recall": 0.8483333333333334,
"f1-score": 0.8568263157087841,
"support": 15000.0
}
}
}
},
"overall": {
"mean_macro_f1": 0.790854835849216,
"mean_accuracy": 0.8479555555555556
}
},
"acsa_no_meta_per_aspect": {
"per_aspect": {
"SIZE": {
"macro_f1": 0.783168541467604,
"accuracy": 0.7674,
"confusion_matrix": [
[
3891,
414,
183
],
[
2322,
5203,
304
],
[
169,
97,
2417
]
],
"report": {
"Not_Mentioned": {
"precision": 0.6096834848010029,
"recall": 0.8669786096256684,
"f1-score": 0.7159153633854646,
"support": 4488.0
},
"Positive": {
"precision": 0.9105705285264263,
"recall": 0.6645804061821433,
"f1-score": 0.7683674222845751,
"support": 7829.0
},
"Negative": {
"precision": 0.8323002754820936,
"recall": 0.900857249347745,
"f1-score": 0.8652228387327725,
"support": 2683.0
},
"accuracy": 0.7674,
"macro avg": {
"precision": 0.7841847629365075,
"recall": 0.8108054217185189,
"f1-score": 0.783168541467604,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8065451857825834,
"recall": 0.7674,
"f1-score": 0.7699979717506622,
"support": 15000.0
}
}
},
"MATERIAL": {
"macro_f1": 0.7558022811037001,
"accuracy": 0.7645333333333333,
"confusion_matrix": [
[
3138,
882,
238
],
[
1764,
6721,
349
],
[
205,
94,
1609
]
],
"report": {
"Not_Mentioned": {
"precision": 0.614450753867241,
"recall": 0.7369657116016909,
"f1-score": 0.6701548318206086,
"support": 4258.0
},
"Positive": {
"precision": 0.8731973496167338,
"recall": 0.7608105048675572,
"f1-score": 0.8131389510616418,
"support": 8834.0
},
"Negative": {
"precision": 0.732695810564663,
"recall": 0.8432914046121593,
"f1-score": 0.7841130604288499,
"support": 1908.0
},
"accuracy": 0.7645333333333333,
"macro avg": {
"precision": 0.7401146380162126,
"recall": 0.7803558736938024,
"f1-score": 0.7558022811037001,
"support": 15000.0
},
"weighted avg": {
"precision": 0.7818760202025543,
"recall": 0.7645333333333333,
"f1-score": 0.7688584324579294,
"support": 15000.0
}
}
},
"QUALITY": {
"macro_f1": 0.7099842589878541,
"accuracy": 0.718,
"confusion_matrix": [
[
5799,
1648,
443
],
[
1797,
4068,
124
],
[
159,
59,
903
]
],
"report": {
"Not_Mentioned": {
"precision": 0.7477756286266924,
"recall": 0.7349809885931559,
"f1-score": 0.7413231064237775,
"support": 7890.0
},
"Positive": {
"precision": 0.7044155844155844,
"recall": 0.6792452830188679,
"f1-score": 0.6916014960897654,
"support": 5989.0
},
"Negative": {
"precision": 0.6142857142857143,
"recall": 0.8055307760927743,
"f1-score": 0.6970281744500193,
"support": 1121.0
},
"accuracy": 0.718,
"macro avg": {
"precision": 0.6888256424426636,
"recall": 0.7399190159015993,
"f1-score": 0.7099842589878541,
"support": 15000.0
},
"weighted avg": {
"precision": 0.7204872620429217,
"recall": 0.718,
"f1-score": 0.7181606168882454,
"support": 15000.0
}
}
},
"APPEARANCE": {
"macro_f1": 0.6901346425171296,
"accuracy": 0.7409333333333333,
"confusion_matrix": [
[
6679,
1462,
429
],
[
1726,
3941,
88
],
[
146,
35,
494
]
],
"report": {
"Not_Mentioned": {
"precision": 0.7810782364635716,
"recall": 0.7793465577596266,
"f1-score": 0.7802114362478827,
"support": 8570.0
},
"Positive": {
"precision": 0.7247149687385068,
"recall": 0.6847958297132928,
"f1-score": 0.7041901188242652,
"support": 5755.0
},
"Negative": {
"precision": 0.4886251236399604,
"recall": 0.7318518518518519,
"f1-score": 0.5860023724792408,
"support": 675.0
},
"accuracy": 0.7409333333333333,
"macro avg": {
"precision": 0.6648061096140129,
"recall": 0.7319980797749238,
"f1-score": 0.6901346425171296,
"support": 15000.0
},
"weighted avg": {
"precision": 0.7462931393359925,
"recall": 0.7409333333333333,
"f1-score": 0.7423051829267658,
"support": 15000.0
}
}
},
"STYLE": {
"macro_f1": 0.6849172239194514,
"accuracy": 0.7234666666666667,
"confusion_matrix": [
[
6020,
1403,
433
],
[
1958,
4249,
158
],
[
150,
46,
583
]
],
"report": {
"Not_Mentioned": {
"precision": 0.7406496062992126,
"recall": 0.7662932790224033,
"f1-score": 0.7532532532532532,
"support": 7856.0
},
"Positive": {
"precision": 0.7457002457002457,
"recall": 0.6675569520816967,
"f1-score": 0.7044682085716655,
"support": 6365.0
},
"Negative": {
"precision": 0.49659284497444633,
"recall": 0.748395378690629,
"f1-score": 0.5970302099334357,
"support": 779.0
},
"accuracy": 0.7234666666666667,
"macro avg": {
"precision": 0.6609808989913015,
"recall": 0.7274152032649096,
"f1-score": 0.6849172239194514,
"support": 15000.0
},
"weighted avg": {
"precision": 0.7301180798135848,
"recall": 0.7234666666666667,
"f1-score": 0.7244389492436237,
"support": 15000.0
}
}
},
"VALUE": {
"macro_f1": 0.7171099329708497,
"accuracy": 0.8563333333333333,
"confusion_matrix": [
[
10588,
815,
378
],
[
786,
1924,
58
],
[
81,
37,
333
]
],
"report": {
"Not_Mentioned": {
"precision": 0.9243125272806635,
"recall": 0.8987352516764281,
"f1-score": 0.9113444654845929,
"support": 11781.0
},
"Positive": {
"precision": 0.6930835734870316,
"recall": 0.6950867052023122,
"f1-score": 0.6940836940836941,
"support": 2768.0
},
"Negative": {
"precision": 0.43302990897269183,
"recall": 0.738359201773836,
"f1-score": 0.5459016393442623,
"support": 451.0
},
"accuracy": 0.8563333333333333,
"macro avg": {
"precision": 0.683475336580129,
"recall": 0.7773937195508588,
"f1-score": 0.7171099329708497,
"support": 15000.0
},
"weighted avg": {
"precision": 0.8668718469501523,
"recall": 0.8563333333333333,
"f1-score": 0.8602649634961278,
"support": 15000.0
}
}
}
},
"overall": {
"mean_macro_f1": 0.7235194801610981,
"mean_accuracy": 0.7617777777777778
}
},
"note": "The Proposed model jointly trains per-aspect heads and an overall sentiment head on the shared fused representation. The overall_head result is the primary overall metric. The aggregated_to_overall result (voting from per-aspect predictions) is included for reference. Baseline 3 vs Proposed isolates the marginal value of metadata cross-attention fusion."
}