Fix gitignore and add LFS

.gitattributes

@@ -1 +1,3 @@
-* text=auto eol=lf
+* text=auto eol=lf
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.csv filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -15,4 +15,6 @@ __pycache__/
 *.png
 *.jpg
 *.json
-*.bat
+*.bat
+__pycache__/
+*.pyc

eval.bat

@@ -0,0 +1,34 @@
+@echo off
+echo ============================================
+echo RECLASS - Full Evaluation Pipeline
+echo ============================================
+
+echo.
+echo [1/6] Evaluate MTL Original
+python src/evaluate.py --mode mtl --dataset original --model_path outputs/best_model_mtl_original.pt
+
+echo.
+echo [2/6] Evaluate MTL Boosted
+python src/evaluate.py --mode mtl --dataset boosted --model_path outputs/best_model_mtl_boosted.pt
+
+echo.
+echo [3/6] Evaluate STL Bug Report
+python src/evaluate.py --mode stl --task bug_report --dataset original --model_path outputs/best_model_stl_bug_report_original.pt
+
+echo.
+echo [4/6] Evaluate STL Feature Request
+python src/evaluate.py --mode stl --task feature_request --dataset original --model_path outputs/best_model_stl_feature_request_original.pt
+
+echo.
+echo [5/6] Evaluate STL Aspect
+python src/evaluate.py --mode stl --task aspect --dataset original --model_path outputs/best_model_stl_aspect_original.pt
+
+echo.
+echo [6/6] Evaluate STL Aspect Sentiment
+python src/evaluate.py --mode stl --task aspect_sentiment --dataset original --model_path outputs/best_model_stl_aspect_sentiment_original.pt
+
+echo.
+echo ============================================
+echo All evaluations complete.
+echo ============================================
+pause

outputs/eval_summary_mtl_mtl_boosted.json

@@ -0,0 +1,186 @@
+{
+    "mode": "mtl",
+    "dataset": "boosted",
+    "task": "all",
+    "model_path": "outputs/best_model_mtl_boosted.pt",
+    "results": {
+        "bug_report": {
+            "macro_f1": 0.9051856266200824,
+            "macro_precision": 0.9016466210687212,
+            "macro_recall": 0.9092221173509517,
+            "confidence": {
+                "overall": 0.9751167297363281,
+                "correct": 0.983551561832428,
+                "incorrect": 0.8863566517829895
+            },
+            "per_class": {
+                "No": {
+                    "precision": 0.9435146443514645,
+                    "recall": 0.9222903885480572,
+                    "f1-score": 0.9327817993795243,
+                    "support": 489.0
+                },
+                "Yes": {
+                    "precision": 0.8597785977859779,
+                    "recall": 0.8961538461538462,
+                    "f1-score": 0.8775894538606404,
+                    "support": 260.0
+                },
+                "accuracy": 0.9132176234979973,
+                "macro avg": {
+                    "precision": 0.9016466210687212,
+                    "recall": 0.9092221173509517,
+                    "f1-score": 0.9051856266200824,
+                    "support": 749.0
+                },
+                "weighted avg": {
+                    "precision": 0.9144473918721233,
+                    "recall": 0.9132176234979973,
+                    "f1-score": 0.9136229077441307,
+                    "support": 749.0
+                }
+            }
+        },
+        "feature_request": {
+            "macro_f1": 0.816421568627451,
+            "macro_precision": 0.8249551363417517,
+            "macro_recall": 0.8084569363081142,
+            "confidence": {
+                "overall": 0.9860208034515381,
+                "correct": 0.9903775453567505,
+                "incorrect": 0.9194391369819641
+            },
+            "per_class": {
+                "No": {
+                    "precision": 0.9633431085043989,
+                    "recall": 0.9690265486725663,
+                    "f1-score": 0.9661764705882353,
+                    "support": 678.0
+                },
+                "Yes": {
+                    "precision": 0.6865671641791045,
+                    "recall": 0.647887323943662,
+                    "f1-score": 0.6666666666666666,
+                    "support": 71.0
+                },
+                "accuracy": 0.9385847797062751,
+                "macro avg": {
+                    "precision": 0.8249551363417517,
+                    "recall": 0.8084569363081142,
+                    "f1-score": 0.816421568627451,
+                    "support": 749.0
+                },
+                "weighted avg": {
+                    "precision": 0.9371066705242975,
+                    "recall": 0.9385847797062751,
+                    "f1-score": 0.9377850205502763,
+                    "support": 749.0
+                }
+            }
+        },
+        "aspect": {
+            "macro_f1": 0.8025782333853203,
+            "macro_precision": 0.8013865970222454,
+            "macro_recall": 0.8071020445874096,
+            "confidence": {
+                "overall": 0.9517425298690796,
+                "correct": 0.9682987332344055,
+                "incorrect": 0.8777835965156555
+            },
+            "per_class": {
+                "App": {
+                    "precision": 0.9329896907216495,
+                    "recall": 0.8418604651162791,
+                    "f1-score": 0.8850855745721271,
+                    "support": 215.0
+                },
+                "Driver": {
+                    "precision": 0.839572192513369,
+                    "recall": 0.8920454545454546,
+                    "f1-score": 0.8650137741046832,
+                    "support": 176.0
+                },
+                "General": {
+                    "precision": 0.8292682926829268,
+                    "recall": 0.7727272727272727,
+                    "f1-score": 0.8,
+                    "support": 88.0
+                },
+                "Payment": {
+                    "precision": 0.7111111111111111,
+                    "recall": 0.8421052631578947,
+                    "f1-score": 0.7710843373493976,
+                    "support": 76.0
+                },
+                "Pricing": {
+                    "precision": 0.8260869565217391,
+                    "recall": 0.8028169014084507,
+                    "f1-score": 0.8142857142857143,
+                    "support": 71.0
+                },
+                "Service": {
+                    "precision": 0.6692913385826772,
+                    "recall": 0.6910569105691057,
+                    "f1-score": 0.68,
+                    "support": 123.0
+                },
+                "accuracy": 0.8170894526034713,
+                "macro avg": {
+                    "precision": 0.8013865970222454,
+                    "recall": 0.8071020445874096,
+                    "f1-score": 0.8025782333853203,
+                    "support": 749.0
+                },
+                "weighted avg": {
+                    "precision": 0.8229006036672395,
+                    "recall": 0.8170894526034713,
+                    "f1-score": 0.8184145769402824,
+                    "support": 749.0
+                }
+            }
+        },
+        "aspect_sentiment": {
+            "macro_f1": 0.6003394063095556,
+            "macro_precision": 0.5894268224819142,
+            "macro_recall": 0.6126239622702458,
+            "confidence": {
+                "overall": 0.9810671806335449,
+                "correct": 0.9874913692474365,
+                "incorrect": 0.8892929553985596
+            },
+            "per_class": {
+                "Positive": {
+                    "precision": 0.8,
+                    "recall": 0.8823529411764706,
+                    "f1-score": 0.8391608391608392,
+                    "support": 136.0
+                },
+                "Neutral": {
+                    "precision": 0.0,
+                    "recall": 0.0,
+                    "f1-score": 0.0,
+                    "support": 6.0
+                },
+                "Negative": {
+                    "precision": 0.9682804674457429,
+                    "recall": 0.9555189456342669,
+                    "f1-score": 0.9618573797678275,
+                    "support": 607.0
+                },
+                "accuracy": 0.9345794392523364,
+                "macro avg": {
+                    "precision": 0.5894268224819142,
+                    "recall": 0.6126239622702458,
+                    "f1-score": 0.6003394063095556,
+                    "support": 749.0
+                },
+                "weighted avg": {
+                    "precision": 0.9299682826963496,
+                    "recall": 0.9345794392523364,
+                    "f1-score": 0.9318735696194198,
+                    "support": 749.0
+                }
+            }
+        }
+    }
+}

outputs/eval_summary_mtl_mtl_original.json

@@ -0,0 +1,186 @@
+{
+    "mode": "mtl",
+    "dataset": "original",
+    "task": "all",
+    "model_path": "outputs/best_model_mtl_original.pt",
+    "results": {
+        "bug_report": {
+            "macro_f1": 0.7833333333333333,
+            "macro_precision": 0.7684555303602922,
+            "macro_recall": 0.8027848820687695,
+            "confidence": {
+                "overall": 0.9568860530853271,
+                "correct": 0.97133868932724,
+                "incorrect": 0.8671128153800964
+            },
+            "per_class": {
+                "No": {
+                    "precision": 0.9319727891156463,
+                    "recall": 0.8954248366013072,
+                    "f1-score": 0.9133333333333333,
+                    "support": 612.0
+                },
+                "Yes": {
+                    "precision": 0.6049382716049383,
+                    "recall": 0.7101449275362319,
+                    "f1-score": 0.6533333333333333,
+                    "support": 138.0
+                },
+                "accuracy": 0.8613333333333333,
+                "macro avg": {
+                    "precision": 0.7684555303602922,
+                    "recall": 0.8027848820687695,
+                    "f1-score": 0.7833333333333333,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.8717984378936761,
+                    "recall": 0.8613333333333333,
+                    "f1-score": 0.8654933333333333,
+                    "support": 750.0
+                }
+            }
+        },
+        "feature_request": {
+            "macro_f1": 0.7632819746470164,
+            "macro_precision": 0.7523338553934565,
+            "macro_recall": 0.7762790132458379,
+            "confidence": {
+                "overall": 0.9604225158691406,
+                "correct": 0.971001923084259,
+                "incorrect": 0.890037477016449
+            },
+            "per_class": {
+                "No": {
+                    "precision": 0.9321486268174475,
+                    "recall": 0.9115323854660348,
+                    "f1-score": 0.9217252396166135,
+                    "support": 633.0
+                },
+                "Yes": {
+                    "precision": 0.5725190839694656,
+                    "recall": 0.6410256410256411,
+                    "f1-score": 0.6048387096774194,
+                    "support": 117.0
+                },
+                "accuracy": 0.8693333333333333,
+                "macro avg": {
+                    "precision": 0.7523338553934565,
+                    "recall": 0.7762790132458379,
+                    "f1-score": 0.7632819746470164,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.8760464181331623,
+                    "recall": 0.8693333333333333,
+                    "f1-score": 0.8722909409460992,
+                    "support": 750.0
+                }
+            }
+        },
+        "aspect": {
+            "macro_f1": 0.717046709402451,
+            "macro_precision": 0.7186594789555653,
+            "macro_recall": 0.7175741849634064,
+            "confidence": {
+                "overall": 0.8980301022529602,
+                "correct": 0.9270597100257874,
+                "incorrect": 0.81709885597229
+            },
+            "per_class": {
+                "App": {
+                    "precision": 0.7875,
+                    "recall": 0.84,
+                    "f1-score": 0.8129032258064516,
+                    "support": 225.0
+                },
+                "Driver": {
+                    "precision": 0.7862595419847328,
+                    "recall": 0.7803030303030303,
+                    "f1-score": 0.7832699619771863,
+                    "support": 132.0
+                },
+                "General": {
+                    "precision": 0.8037383177570093,
+                    "recall": 0.688,
+                    "f1-score": 0.7413793103448276,
+                    "support": 125.0
+                },
+                "Payment": {
+                    "precision": 0.6176470588235294,
+                    "recall": 0.6363636363636364,
+                    "f1-score": 0.6268656716417911,
+                    "support": 33.0
+                },
+                "Pricing": {
+                    "precision": 0.6986301369863014,
+                    "recall": 0.75,
+                    "f1-score": 0.723404255319149,
+                    "support": 68.0
+                },
+                "Service": {
+                    "precision": 0.6181818181818182,
+                    "recall": 0.6107784431137725,
+                    "f1-score": 0.6144578313253012,
+                    "support": 167.0
+                },
+                "accuracy": 0.736,
+                "macro avg": {
+                    "precision": 0.7186594789555653,
+                    "recall": 0.7175741849634064,
+                    "f1-score": 0.717046709402451,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.7367554868722928,
+                    "recall": 0.736,
+                    "f1-score": 0.7352797185836668,
+                    "support": 750.0
+                }
+            }
+        },
+        "aspect_sentiment": {
+            "macro_f1": 0.7574652640875613,
+            "macro_precision": 0.7713119380138362,
+            "macro_recall": 0.7471910012842655,
+            "confidence": {
+                "overall": 0.9650789499282837,
+                "correct": 0.9762944579124451,
+                "incorrect": 0.8488451242446899
+            },
+            "per_class": {
+                "Positive": {
+                    "precision": 0.9387755102040817,
+                    "recall": 0.9533678756476683,
+                    "f1-score": 0.9460154241645244,
+                    "support": 386.0
+                },
+                "Neutral": {
+                    "precision": 0.45161290322580644,
+                    "recall": 0.358974358974359,
+                    "f1-score": 0.4,
+                    "support": 39.0
+                },
+                "Negative": {
+                    "precision": 0.9235474006116208,
+                    "recall": 0.9292307692307692,
+                    "f1-score": 0.9263803680981595,
+                    "support": 325.0
+                },
+                "accuracy": 0.912,
+                "macro avg": {
+                    "precision": 0.7713119380138362,
+                    "recall": 0.7471910012842655,
+                    "f1-score": 0.7574652640875613,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.9068442071511451,
+                    "recall": 0.912,
+                    "f1-score": 0.9091140978125444,
+                    "support": 750.0
+                }
+            }
+        }
+    }
+}

outputs/eval_summary_stl_aspect_original.json

@@ -0,0 +1,69 @@
+{
+    "mode": "stl",
+    "dataset": "original",
+    "task": "aspect",
+    "model_path": "outputs/best_model_stl_aspect_original.pt",
+    "results": {
+        "aspect": {
+            "macro_f1": 0.6941267385341167,
+            "macro_precision": 0.6926366440012734,
+            "macro_recall": 0.7177510694594175,
+            "confidence": {
+                "overall": 0.8459477424621582,
+                "correct": 0.8876724243164062,
+                "incorrect": 0.7407546639442444
+            },
+            "per_class": {
+                "App": {
+                    "precision": 0.7973568281938326,
+                    "recall": 0.8044444444444444,
+                    "f1-score": 0.8008849557522124,
+                    "support": 225.0
+                },
+                "Driver": {
+                    "precision": 0.7172413793103448,
+                    "recall": 0.7878787878787878,
+                    "f1-score": 0.7509025270758123,
+                    "support": 132.0
+                },
+                "General": {
+                    "precision": 0.7619047619047619,
+                    "recall": 0.64,
+                    "f1-score": 0.6956521739130435,
+                    "support": 125.0
+                },
+                "Payment": {
+                    "precision": 0.625,
+                    "recall": 0.6060606060606061,
+                    "f1-score": 0.6153846153846154,
+                    "support": 33.0
+                },
+                "Pricing": {
+                    "precision": 0.5614035087719298,
+                    "recall": 0.9411764705882353,
+                    "f1-score": 0.7032967032967034,
+                    "support": 68.0
+                },
+                "Service": {
+                    "precision": 0.6929133858267716,
+                    "recall": 0.5269461077844312,
+                    "f1-score": 0.5986394557823129,
+                    "support": 167.0
+                },
+                "accuracy": 0.716,
+                "macro avg": {
+                    "precision": 0.6926366440012734,
+                    "recall": 0.7177510694594175,
+                    "f1-score": 0.6941267385341167,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.7251149569069801,
+                    "recall": 0.716,
+                    "f1-score": 0.7125059034731999,
+                    "support": 750.0
+                }
+            }
+        }
+    }
+}

outputs/eval_summary_stl_aspect_sentiment_original.json

@@ -0,0 +1,51 @@
+{
+    "mode": "stl",
+    "dataset": "original",
+    "task": "aspect_sentiment",
+    "model_path": "outputs/best_model_stl_aspect_sentiment_original.pt",
+    "results": {
+        "aspect_sentiment": {
+            "macro_f1": 0.7856154710827541,
+            "macro_precision": 0.7905014749262537,
+            "macro_recall": 0.7816500597847748,
+            "confidence": {
+                "overall": 0.9811832904815674,
+                "correct": 0.987269401550293,
+                "incorrect": 0.9148151278495789
+            },
+            "per_class": {
+                "Positive": {
+                    "precision": 0.96,
+                    "recall": 0.9326424870466321,
+                    "f1-score": 0.9461235216819974,
+                    "support": 386.0
+                },
+                "Neutral": {
+                    "precision": 0.5,
+                    "recall": 0.46153846153846156,
+                    "f1-score": 0.48,
+                    "support": 39.0
+                },
+                "Negative": {
+                    "precision": 0.911504424778761,
+                    "recall": 0.9507692307692308,
+                    "f1-score": 0.9307228915662651,
+                    "support": 325.0
+                },
+                "accuracy": 0.916,
+                "macro avg": {
+                    "precision": 0.7905014749262537,
+                    "recall": 0.7816500597847748,
+                    "f1-score": 0.7856154710827541,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.9150652507374631,
+                    "recall": 0.916,
+                    "f1-score": 0.9152114921710495,
+                    "support": 750.0
+                }
+            }
+        }
+    }
+}

outputs/eval_summary_stl_bug_report_original.json

@@ -0,0 +1,45 @@
+{
+    "mode": "stl",
+    "dataset": "original",
+    "task": "bug_report",
+    "model_path": "outputs/best_model_stl_bug_report_original.pt",
+    "results": {
+        "bug_report": {
+            "macro_f1": 0.7845034000574658,
+            "macro_precision": 0.7627192982456141,
+            "macro_recall": 0.8191602728047741,
+            "confidence": {
+                "overall": 0.9904617667198181,
+                "correct": 0.9926525950431824,
+                "incorrect": 0.9774385094642639
+            },
+            "per_class": {
+                "No": {
+                    "precision": 0.9421052631578948,
+                    "recall": 0.8774509803921569,
+                    "f1-score": 0.9086294416243654,
+                    "support": 612.0
+                },
+                "Yes": {
+                    "precision": 0.5833333333333334,
+                    "recall": 0.7608695652173914,
+                    "f1-score": 0.660377358490566,
+                    "support": 138.0
+                },
+                "accuracy": 0.856,
+                "macro avg": {
+                    "precision": 0.7627192982456141,
+                    "recall": 0.8191602728047741,
+                    "f1-score": 0.7845034000574658,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.8760912280701755,
+                    "recall": 0.856,
+                    "f1-score": 0.8629510583277463,
+                    "support": 750.0
+                }
+            }
+        }
+    }
+}

outputs/eval_summary_stl_feature_request_original.json

@@ -0,0 +1,45 @@
+{
+    "mode": "stl",
+    "dataset": "original",
+    "task": "feature_request",
+    "model_path": "outputs/best_model_stl_feature_request_original.pt",
+    "results": {
+        "feature_request": {
+            "macro_f1": 0.7419244916163044,
+            "macro_precision": 0.739484126984127,
+            "macro_recall": 0.7444606473042492,
+            "confidence": {
+                "overall": 0.9515063762664795,
+                "correct": 0.9664607644081116,
+                "incorrect": 0.8575695753097534
+            },
+            "per_class": {
+                "No": {
+                    "precision": 0.9206349206349206,
+                    "recall": 0.9162717219589257,
+                    "f1-score": 0.9184481393507522,
+                    "support": 633.0
+                },
+                "Yes": {
+                    "precision": 0.5583333333333333,
+                    "recall": 0.5726495726495726,
+                    "f1-score": 0.5654008438818565,
+                    "support": 117.0
+                },
+                "accuracy": 0.8626666666666667,
+                "macro avg": {
+                    "precision": 0.739484126984127,
+                    "recall": 0.7444606473042492,
+                    "f1-score": 0.7419244916163044,
+                    "support": 750.0
+                },
+                "weighted avg": {
+                    "precision": 0.864115873015873,
+                    "recall": 0.8626666666666667,
+                    "f1-score": 0.8633727612576044,
+                    "support": 750.0
+                }
+            }
+        }
+    }
+}

run.bat

@@ -0,0 +1,34 @@
+@echo off
+echo ============================================
+echo RECLASS - Full Training Pipeline
+echo ============================================
+
+echo.
+echo [1/6] MTL Original
+python src/train.py --mode mtl --dataset original --epochs 15 --patience 3
+
+echo.
+echo [2/6] MTL Boosted
+python src/train.py --mode mtl --dataset boosted --epochs 15 --patience 3
+
+echo.
+echo [3/6] STL Bug Report
+python src/train.py --mode stl --task bug_report --dataset original --epochs 15 --patience 3
+
+echo.
+echo [4/6] STL Feature Request
+python src/train.py --mode stl --task feature_request --dataset original --epochs 15 --patience 3
+
+echo.
+echo [5/6] STL Aspect
+python src/train.py --mode stl --task aspect --dataset original --epochs 15 --patience 3
+
+echo.
+echo [6/6] STL Aspect Sentiment
+python src/train.py --mode stl --task aspect_sentiment --dataset original --epochs 15 --patience 3
+
+echo.
+echo ============================================
+echo All training runs complete.
+echo ============================================
+pause