NYXAR BiLSTM Sentiment Model
Model Description
A deep learning sentiment classification model developed for the NYXAR AI Intelligence & Observability Platform.
The model uses a Bidirectional Long Short-Term Memory (BiLSTM) neural network to capture contextual information from text sequences and predict sentiment across three classes:
- Positive
- Neutral
- Negative
Framework:
- TensorFlow
- Keras
- BiLSTM Architecture
Language:
- English
License:
- MIT
Training Data
Dataset:
- SetFit/amazon_reviews_multi_en
The dataset contains English Amazon product reviews used for sentiment classification.
Intended Use
This model is designed for:
- Customer feedback analysis
- Product review monitoring
- Support ticket intelligence
- Sentiment trend analysis
Limitations
The model may struggle with:
- Sarcasm
- Irony
- Ambiguous sentiment
- Domain-specific terminology not represented in the training data
Predictions should be used as supporting signals rather than business-critical decisions.
Performance
Metrics obtained during evaluation:
- Accuracy: 72.82%
- Precision: 70.55%
- Recall: 72.82%
- F1 Score: 71.29%
Usage
import joblib
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.sequence import pad_sequences
model = load_model("bilstm_model_v1.keras")
tokenizer = joblib.load("bilstm_tokenizer_v1.pkl")
max_len = joblib.load("max_len_v1.pkl")
text = ["The product exceeded expectations."]
sequence = tokenizer.texts_to_sequences(text)
padded = pad_sequences(sequence, maxlen=max_len)
prediction = model.predict(padded)
- Downloads last month
- 101