A Study on Deep Learning Based Anomaly Detection Technique in Natural Language Processing Textual Data

Anomaly detection is used to identify outliers in datasets. Anomaly detection has been in use for the past four decades for the purpose of extracting out the anomalous parts from data has been used for decades to identify and extract anomalous components from data. However, identification and study of the pattern of outliers can come in handy when making future business decisions, or improve the precision of the learning model. In detail, the current demanding of anomaly detection is investigated, with some principles of anomaly detection are also outlined and some applications which are highlighted. The following is a research paper current work proposing the application of supervised as well as unsupervised learning for effecting the prognosis of the anomaly detection. We also utilize NLP (Natural Language Processing) and for this, it uses the word Embeddings technique which treats words as highdimensional space. Here they pick up the semantic level, and words, which mean similar things, will be close in the vector space. The process begins with clean-up of the dataset and moving the features that are of value out of the pre-processing step using sophisticated techniques like word embeddings and sentiment analysis. The proposed methodologies demonstrate superior detection performance compared to existing methods in the literature. Anomaly detection is of course the most relevant to our work, used most notably in fraud detection, semantic analysis and text data error correction.