The purpose of the proposed pipeline is to classify Amazon product reviews as normal or anomalous based on their content, using pretrained MPNet transformer for encoding and DAEF autoencoder for anomaly detection.
The pipeline includes three main modules: Text encoding, Anomaly detection, and Explainability, each serving a specific purpose in the classification process.
Collection
[
|
...
]