4/25/25, 9:17 AM DL5.
ipynb - Colab
import networkx as nx
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.sequence import pad_sequences
from tensorflow.keras.preprocessing.text import Tokenizer
import pickle
import tensorflow as tf
from tensorflow.keras.datasets import imdb
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Embedding, LSTM, Dense
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix
vocab_size = 10000
max_length = 200
embedding_dim = 128
lstm_units = 128
epochs = 5
batch_size = 32
(X_train, y_train), (X_test, y_test) = imdb.load_data(num_words=vocab_size)
X_train = pad_sequences(X_train, maxlen=max_length)
X_test = pad_sequences(X_test, maxlen=max_length)
add Code add Text
model = Sequential()
model.add(Embedding(input_dim=vocab_size, output_dim=embedding_dim, input_length=max_length))
model.add(LSTM(units=lstm_units))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
model.summary()
c:\Users\Admin\anaconda3\envs\venv\Lib\site-packages\keras\src\layers\core\embedding.py:90: UserWarning: Argument `input_length` is
warnings.warn(
Model: "sequential"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type) ┃ Output Shape ┃ Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ embedding (Embedding) │ ? │ 0 (unbuilt) │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ lstm (LSTM) │ ? │ 0 (unbuilt) │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense (Dense) │ ? │ 0 (unbuilt) │
└─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 0 (0.00 B)
Trainable params: 0 (0 00 B)
history = model.fit(X_train,y_train,epochs=epochs,batch_size=batch_size,validation_split=0.2)
Epoch 1/5
625/625 ━━━━━━━━━━━━━━━━━━━━ 87s 134ms/step - accuracy: 0.7330 - loss: 0.5250 - val_accuracy: 0.8274 - val_loss: 0.3947
Epoch 2/5
625/625 ━━━━━━━━━━━━━━━━━━━━ 69s 109ms/step - accuracy: 0.8840 - loss: 0.2864 - val_accuracy: 0.8682 - val_loss: 0.3159
Epoch 3/5
625/625 ━━━━━━━━━━━━━━━━━━━━ 81s 130ms/step - accuracy: 0.9321 - loss: 0.1883 - val_accuracy: 0.8528 - val_loss: 0.4546
Epoch 4/5
625/625 ━━━━━━━━━━━━━━━━━━━━ 73s 117ms/step - accuracy: 0.9518 - loss: 0.1355 - val_accuracy: 0.8544 - val_loss: 0.3734
Epoch 5/5
625/625 ━━━━━━━━━━━━━━━━━━━━ 66s 106ms/step - accuracy: 0.9647 - loss: 0.1005 - val_accuracy: 0.8538 - val_loss: 0.4103
test_loss, test_accuracy = model.evaluate(X_test,y_test)
print(f"Test Accuracy:{test_accuracy:.4f}, Test Loss:{test_loss:.4f}")
y_pred_probs = model.predict(X_test)
y_pred_classes = (y_pred_probs > 0.5).astype(int)
conf_matrix = confusion_matrix(y_test, y_pred_classes)
782/782 ━━━━━━━━━━━━━━━━━━━━ 40s 52ms/step - accuracy: 0.8491 - loss: 0.4451
Test Accuracy:0.8516, Test Loss:0.4352
782/782 ━━━━━━━━━━━━━━━━━━━━ 49s 61ms/step
plt.figure(figsize = (8, 6))
sns.heatmap(conf_matrix, annot = True, cmap = 'Blues', fmt = 'd', xticklabels = ['Negative', 'Positive'], yticklabels = ['Negative', 'Po
plt.xlabel('Predicted')
plt.ylabel('Actual')
plt.title('Confusion Matrix')
plt.show()
https://colab.research.google.com/drive/1KG8n08q7-0gBxcBoeeqvwrV7v20RXmZB#printMode=true 1/3
4/25/25, 9:17 AM DL5.ipynb - Colab
TN, FP, FN, TP = conf_matrix.ravel()
accuracy_cm = (TN + TP) / (TN + TP + FN + FP)
precision_cm = TP / (TP + FP)
recall_cm = TP / (TP + FN)
f1_score_cm = 2 * (precision_cm * recall_cm) / (precision_cm + recall_cm)
print(f"Accuracy: {accuracy_cm:.4f}")
print(f"Precision: {precision_cm:.4f}")
print(f"Recall: {recall_cm:.4f}")
print(f"F1 Score: {f1_score_cm:.4f}")
Accuracy: 0.8516
Precision: 0.8707
Recall: 0.8259
F1 Score: 0.8477
model.save('sentiment_rnn_model.h5')
WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is c
word_index = imdb.get_word_index()
tokenizer = Tokenizer(num_words=vocab_size)
tokenizer.word_index = word_index
with open('tokenizer.pickle','wb') as handle:
pickle.dump(tokenizer,handle)
print("Model and tokenizer saved successfully!")
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/imdb_word_index.json
1641221/1641221 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
Model and tokenizer saved successfully!
def perform_sentiment_analysis(graph,model_path,tokenizer_path,max_length):
"""
Perform sentiment analysis on a network graph
using a pre-trained RNN model.
Parameters:
- graph (nx.Graph): NetworkX graph with 'text'
attributes on nodes.
- model_path (str): Path to the pre-trained RN
N model file (e.g., 'sentiment_rnn_model.h5')
- tokenizer_path (str): Path to the saved toke
nizer file (e.g., 'tokenizer.pickle').
- max_length (int): Maximum sequence length fo
r padding (must match the model's training).
Returns:
- nx.Graph: The input graph with 'sentiment' a
ttributes added to nodes.
"""
texts = [data['text'] for _,data in graph.nodes(data=True)]
https://colab.research.google.com/drive/1KG8n08q7-0gBxcBoeeqvwrV7v20RXmZB#printMode=true 2/3
4/25/25, 9:17 AM DL5.ipynb - Colab
if not texts:
raise ValueError("Graph is empty or 'text' attribute is missing.")
with open(tokenizer_path,'rb') as handle:
tokenizer = pickle.load(handle)
sequences = tokenizer.texts_to_sequences(texts)
padded_sequences = pad_sequences(sequences,maxlen=max_length)
model = load_model(model_path)
predictions = model.predict(padded_sequences)
sentiments = ['positive' if p > 0.5 else 'negative' for p in predictions.flatten()]
for node, sentiment in zip(graph.nodes(), sentiments):
graph.nodes[node]['sentiment'] = sentiment
return graph
if __name__ == "__main__":
# Create a sample graph
G = nx.Graph()
G.add_node(1, text="This is terrible.")
G.add_node(2, text="It's absolutely amazing.")
# Specify paths and parameters (replace with actual paths)
model_path = 'sentiment_rnn_model.h5'
tokenizer_path = 'tokenizer.pickle'
max_length = 200 # Must match the model's training configuration
try:
# Perform sentiment analysis
G = perform_sentiment_analysis(G, model_path, tokenizer_path, max_length)
except ValueError as e:
print(f"Error: {e}")
# Visualize the sentiment analysis results
pos_count = sum([1 for _, data in G.nodes(data=True) if data['sentiment'] == 'positive'])
neg_count = sum([1 for _, data in G.nodes(data=True) if data['sentiment'] == 'negative'])
total_count = len(G.nodes())
pos_percent = (pos_count / total_count) * 100
neg_percent = (neg_count / total_count) * 100
print(f"Positive Sentiment: {pos_percent:.2f}%")
print(f"Negative Sentiment: {neg_percent:.2f}%")
WARNING:absl:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until y
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 233ms/step
Positive Sentiment: 0.00%
Negative Sentiment: 100.00%
Start coding or generate with AI.
https://colab.research.google.com/drive/1KG8n08q7-0gBxcBoeeqvwrV7v20RXmZB#printMode=true 3/3
