Integrated Recommendation and Classification System for Medical Data Using Content-Based Filtering and LSTM Neural Networks

In today’s world people are using recommendation system everywhere. In this proposed work, investigates system combines content-based filtering using TF-IDF vectorization and nearest neighbors algorithm for recommendation, along with a Bidirectional LSTM (Long Short-Term Memory) neural network for classification tasks. The proposed hybrid filtering approach aims to enhance personalized recommendation generation by leveraging the strengths of both content-based and collaborative filtering techniques. Specifically, TF-IDF vectorization is employed to extract meaningful features from textual data, capturing the semantic similarities between items. The Nearest Neighbors algorithm, utilizing cosine similarity as the distance metric, identifies the most similar items based on the TF-IDF vectors. the classification system utilizes the LSTM model to predict categories based on the description text. We evaluate the system’s performance using various metrics, including f1 Score, precision, recall, and accuracy, ensuring its robustness and efficacy. Bi LSTM models provided an accuracy of 95%, This combined approach offers a more robust and accurate recommendation system, addressing the limitations of traditional methods. Experimental evaluations demonstrate the effectiveness and superiority of the hybrid filtering model in generating personalized recommendations based on user queries. The findings highlight the potential of integrating diverse techniques in recommendation systems to achieve enhanced user satisfaction and engagement.