len_q1 vs len_q2 | diff_len |
---|---|
![]() |
![]() |
cosine_distance vs common_words | fuzz_partial_ratio vs common_words |
---|---|
![]() |
![]() |
S.N. | Supervised Machine Learning | Accuracy |
---|---|---|
1 | Random Forest | 0.7235 |
2 | K Nearest Neighbors | 0.7104 |
3 | Logistic Regression | 0.6680 |
S.N. | Learning Rate | Batch Size | Training Accuracy |
---|---|---|---|
1 | 0.01 | 30 | 0.6264 |
2 | 0.005 | 30 | 0.6938 |
3 | 0.001 | 30 | 0.7256 |
Training Loss | Training Accuracy | ROC |
---|---|---|
![]() |
![]() |
![]() |
Predicted | |||
---|---|---|---|
0 | 1 | ||
Actual | 0 | 36462 | 12792 |
0 | 8954 | 20455 |
Measure | Value | Derivations |
---|---|---|
Sensitivity | 0.6955 | TPR = TP / (TP + FN) |
Specificity | 0.7403 | SPC = TN / (FP + TN) |
Precision | 0.6152 | PPV = TP / (TP + FP) |
Negative Predictive Value | 0.8028 | NPV = TN / (TN + FN) |
False Positive Rate | 0.2597 | FPR = FP / (FP + TN) |
False Discovery Rate | 0.3848 | FDR = FP / (FP + TP) |
False Negative Rate | 0.3045 | FNR = FN / (FN + TP) |
Accuracy | 0.7236 | ACC = (TP + TN) / (P + N) |
F1 Score | 0.6529 | F1 = 2TP / (2TP + FP + FN) |
pip3 install numpy
pip3 install fuzzywuzzy
pip3 install gensim
pip3 install python-Levenshtein
pip3 install sklearn
pip3 install pyemd
pip3 install keras
pip3 install tensorflow
Download Models from https://drive.google.com/open?id=1YgibRxIBRPDBvrPPstxkInnNKc6M5lFc & Extract to ./semantic-question-matching/flask_interface/
cd ./semantic-question-matching/flask_interface
export FLASK_APP=app.py
flask run --without-threads