bernard-ng/drc-ners-nlp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: add more baseline expirements

Browse Source
This commit is contained in:
bernard-ng
2025-09-21 00:06:01 +02:00
parent e41b15a863
commit 83d21c640b
3 changed files with 278 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+40 -5
View File
@@ -65,7 +65,7 @@ stages:
python main.py --env development
```
## NER Processing
## NER Processing (Optional)
This project implements a custom named entity recognition (NER) pipeline tailored for Congolese names.
Its main objective is to accurately identify and tag the different components of a Congolese name,
@@ -75,7 +75,7 @@ specifically distinguishing between the native part and the surname.
python ner.py --env development
```
Once you've built and train the NER model you can use it to annotate **CoMPOSE** name in the original dataset
Once you've built and train the NER model you can use it to annotate **COMPOSE** name in the original dataset
**Running the Pipeline with NER Annotation**
```yaml
@@ -104,20 +104,55 @@ you can define model features, training parameters, and evaluation metrics in th
**Running Experiments**
```bash
# bigru
python train.py --name="bigru" --type="baseline" --env="development"
python train.py --name="cnn" --type="baseline" --env="development"
python train.py --name="lightgbm" --type="baseline" --env="development"
python train.py --name="bigru_native" --type="baseline" --env="development"
python train.py --name="bigru_surname" --type="baseline" --env="development"
python train.py --name="logistic_regression_fullname" --type="baseline" --env="development"
# cnn
python train.py --name="cnn" --type="baseline" --env="development"
python train.py --name="cnn_native" --type="baseline" --env="development"
python train.py --name="cnn_surname" --type="baseline" --env="development"
# lightgbm
python train.py --name="lightgbm" --type="baseline" --env="development"
python train.py --name="lightgbm_native" --type="baseline" --env="development"
python train.py --name="lightgbm_surname" --type="baseline" --env="development"
# logistic regression
python train.py --name="logistic_regression" --type="baseline" --env="development"
python train.py --name="logistic_regression_native" --type="baseline" --env="development"
python train.py --name="logistic_regression_surname" --type="baseline" --env="development"
# lstm
python train.py --name="lstm" --type="baseline" --env="development"
python train.py --name="lstm_native" --type="baseline" --env="development"
python train.py --name="lstm_surname" --type="baseline" --env="development"
# random forest
python train.py --name="random_forest" --type="baseline" --env="development"
python train.py --name="random_forest_native" --type="baseline" --env="development"
python train.py --name="random_forest_surname" --type="baseline" --env="development"
# svm
python train.py --name="svm" --type="baseline" --env="development"
python train.py --name="svm_native" --type="baseline" --env="development"
python train.py --name="svm_surname" --type="baseline" --env="development"
# naive bayes
python train.py --name="naive_bayes" --type="baseline" --env="development"
python train.py --name="naive_bayes_native" --type="baseline" --env="development"
python train.py --name="naive_bayes_surname" --type="baseline" --env="development"
# transformer
python train.py --name="transformer" --type="baseline" --env="development"
python train.py --name="transformer_native" --type="baseline" --env="development"
python train.py --name="transformer_surname" --type="baseline" --env="development"
# xgboost
python train.py --name="xgboost" --type="baseline" --env="development"
python train.py --name="xgboost_native" --type="baseline" --env="development"
python train.py --name="xgboost_surname" --type="baseline" --env="development"
```
## Web Interface
config/research_templates.yaml
+236 -1
View File
@@ -1,4 +1,5 @@
baseline_experiments:
# BiGRU Models
- name: "bigru"
description: "Baseline BiGRU with full name features"
model_type: "bigru"
@@ -10,6 +11,29 @@ baseline_experiments:
batch_size: 32
tags: [ "baseline", "neural", "bigru" ]
- name: "bigru_native"
description: "Baseline BiGRU with native name features"
model_type: "bigru"
features: [ "native_name" ]
model_params:
embedding_dim: 64
gru_units: 32
epochs: 2
batch_size: 32
tags: [ "baseline", "neural", "bigru", "native" ]
- name: "bigru_surname"
description: "Baseline BiGRU with surname features"
model_type: "bigru"
features: [ "surname" ]
model_params:
embedding_dim: 64
gru_units: 32
epochs: 2
batch_size: 32
tags: [ "baseline", "neural", "bigru", "surname" ]
## CNN Models
- name: "cnn"
description: "Baseline CNN with character patterns"
model_type: "cnn"
@@ -23,6 +47,33 @@ baseline_experiments:
batch_size: 32
tags: [ "baseline", "neural", "cnn" ]
- name: "cnn_native"
description: "Baseline CNN with native name character patterns"
model_type: "cnn"
features: [ "native_name" ]
model_params:
embedding_dim: 64
filters: 64
kernel_size: 3
dropout: 0.5
epochs: 2
batch_size: 32
tags: [ "baseline", "neural", "cnn", "native" ]
- name: "cnn_surname"
description: "Baseline CNN with surname character patterns"
model_type: "cnn"
features: [ "surname" ]
model_params:
embedding_dim: 64
filters: 64
kernel_size: 3
dropout: 0.5
epochs: 2
batch_size: 32
tags: [ "baseline", "neural", "cnn", "surname" ]
## Ensemble Models
- name: "ensemble"
description: "Baseline Ensemble with multiple models"
model_type: "ensemble"
@@ -33,6 +84,27 @@ baseline_experiments:
cv_folds: 5
tags: [ "baseline", "ensemble" ]
- name: "ensemble_native"
description: "Baseline Ensemble with native name"
model_type: "ensemble"
features: [ "native_name" ]
model_params:
base_models: [ "logistic_regression", "random_forest", "xgboost" ]
voting: "soft"
cv_folds: 5
tags: [ "baseline", "ensemble", "native" ]
- name: "ensemble_surname"
description: "Baseline Ensemble with surname"
model_type: "ensemble"
features: [ "surname" ]
model_params:
base_models: [ "logistic_regression", "random_forest", "xgboost" ]
voting: "soft"
cv_folds: 5
tags: [ "baseline", "ensemble", "surname" ]
# LightGBM Models
- name: "lightgbm"
description: "Baseline LightGBM with engineered features"
model_type: "lightgbm"
@@ -46,7 +118,34 @@ baseline_experiments:
colsample_bytree: 0.8
tags: [ "baseline", "lightgbm" ]
- name: "logistic_regression_fullname"
- name: "lightgbm_native"
description: "Baseline LightGBM with native name features"
model_type: "lightgbm"
features: [ "native_name" ]
model_params:
n_estimators: 100
max_depth: -1
learning_rate: 0.1
num_leaves: 31
subsample: 0.8
colsample_bytree: 0.8
tags: [ "baseline", "lightgbm", "native" ]
- name: "lightgbm_surname"
description: "Baseline LightGBM with surname features"
model_type: "lightgbm"
features: [ "surname" ]
model_params:
n_estimators: 100
max_depth: -1
learning_rate: 0.1
num_leaves: 31
subsample: 0.8
colsample_bytree: 0.8
tags: [ "baseline", "lightgbm", "surname" ]
# Logistic Regression Models
- name: "logistic_regression"
description: "Baseline logistic regression with full name"
model_type: "logistic_regression"
features: [ "full_name" ]
@@ -70,6 +169,7 @@ baseline_experiments:
max_features: 5000
tags: [ "baseline", "logistic_regression", "surname" ]
# LSTM Models
- name: "lstm"
description: "Baseline LSTM with full name features"
model_type: "lstm"
@@ -81,6 +181,29 @@ baseline_experiments:
batch_size: 64
tags: [ "baseline", "neural", "lstm" ]
- name: "lstm_native"
description: "Baseline LSTM with native name features"
model_type: "lstm"
features: [ "native_name" ]
model_params:
embedding_dim: 128
lstm_units: 64
epochs: 2
batch_size: 64
tags: [ "baseline", "neural", "lstm", "native" ]
- name: "lstm_surname"
description: "Baseline LSTM with surname features"
model_type: "lstm"
features: [ "surname" ]
model_params:
embedding_dim: 128
lstm_units: 64
epochs: 2
batch_size: 64
tags: [ "baseline", "neural", "lstm", "surname" ]
# Naive Bayes Models
- name: "naive_bayes"
description: "Baseline Naive Bayes with full name features"
model_type: "naive_bayes"
@@ -89,6 +212,23 @@ baseline_experiments:
max_features: 5000
tags: [ "baseline", "naive_bayes" ]
- name: "naive_bayes_native"
description: "Baseline Naive Bayes with native name features"
model_type: "naive_bayes"
features: [ "native_name" ]
model_params:
max_features: 5000
tags: [ "baseline", "naive_bayes", "native" ]
- name: "naive_bayes_surname"
description: "Baseline Naive Bayes with surname features"
model_type: "naive_bayes"
features: [ "surname" ]
model_params:
max_features: 5000
tags: [ "baseline", "naive_bayes", "surname" ]
# Random Forest Models
- name: "random_forest"
description: "Baseline Random Forest with engineered features"
model_type: "random_forest"
@@ -100,6 +240,29 @@ baseline_experiments:
min_samples_leaf: 1
tags: [ "baseline", "random_forest", "engineered" ]
- name: "random_forest_native"
description: "Baseline Random Forest with native name engineered features"
model_type: "random_forest"
features: [ "native_name" ]
model_params:
n_estimators: 100
max_depth: 10
min_samples_split: 2
min_samples_leaf: 1
tags: [ "baseline", "random_forest", "engineered", "native" ]
- name: "random_forest_surname"
description: "Baseline Random Forest with surname engineered features"
model_type: "random_forest"
features: [ "surname" ]
model_params:
n_estimators: 100
max_depth: 10
min_samples_split: 2
min_samples_leaf: 1
tags: [ "baseline", "random_forest", "engineered", "surname" ]
# SVM Models
- name: "svm"
description: "Baseline SVM with full name features"
model_type: "svm"
@@ -111,6 +274,29 @@ baseline_experiments:
max_features: 5000
tags: [ "baseline", "svm" ]
- name: "svm_native"
description: "Baseline SVM with native name features"
model_type: "svm"
features: [ "native_name" ]
model_params:
C: 1.0
kernel: "rbf"
ngram_range: [ 2, 4 ]
max_features: 5000
tags: [ "baseline", "svm", "native" ]
- name: "svm_surname"
description: "Baseline SVM with surname features"
model_type: "svm"
features: [ "surname" ]
model_params:
C: 1.0
kernel: "rbf"
ngram_range: [ 2, 4 ]
max_features: 5000
tags: [ "baseline", "svm", "surname" ]
# Transformer Models
- name: "transformer"
description: "Baseline Transformer with attention mechanism"
model_type: "transformer"
@@ -123,6 +309,31 @@ baseline_experiments:
batch_size: 64
tags: [ "baseline", "neural", "transformer" ]
- name: "transformer_native"
description: "Baseline Transformer with native name attention mechanism"
model_type: "transformer"
features: [ "native_name" ]
model_params:
embedding_dim: 128
num_heads: 4
num_layers: 2
epochs: 2
batch_size: 64
tags: [ "baseline", "neural", "transformer", "native" ]
- name: "transformer_surname"
description: "Baseline Transformer with surname attention mechanism"
model_type: "transformer"
features: [ "surname" ]
model_params:
embedding_dim: 128
num_heads: 4
num_layers: 2
epochs: 2
batch_size: 64
tags: [ "baseline", "neural", "transformer", "surname" ]
# XGBoost Models
- name: "xgboost"
description: "Baseline XGBoost with engineered features"
model_type: "xgboost"
@@ -135,6 +346,30 @@ baseline_experiments:
colsample_bytree: 0.8
tags: [ "baseline", "xgboost" ]
- name: "xgboost_native"
description: "Baseline XGBoost with native name engineered features"
model_type: "xgboost"
features: [ "native_name" ]
model_params:
n_estimators: 100
max_depth: 6
learning_rate: 0.1
subsample: 0.8
colsample_bytree: 0.8
tags: [ "baseline", "xgboost", "native" ]
- name: "xgboost_surname"
description: "Baseline XGBoost with surname engineered features"
model_type: "xgboost"
features: [ "surname" ]
model_params:
n_estimators: 100
max_depth: 6
learning_rate: 0.1
subsample: 0.8
colsample_bytree: 0.8
tags: [ "baseline", "xgboost", "surname" ]
# Advanced Experiments Configuration
advanced_experiments:
research/models/cnn_model.py
+2 -3
View File
@@ -12,6 +12,8 @@ from tensorflow.keras.layers import (
SpatialDropout1D,
)
from tensorflow.keras.models import Sequential
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
from research.neural_network_model import NeuralNetworkModel
@@ -64,9 +66,6 @@ class CNNModel(NeuralNetworkModel):
def prepare_features(self, X: pd.DataFrame) -> np.ndarray:
"""Prepare sequences for CNN using extracted features"""
# X here contains the features already extracted by FeatureExtractor
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
# Get text data from extracted features - use character level for CNN
text_data = self._collect_text_corpus(X)