Implementasi Sistem Rekomendasi Laptop Berdasarkan Kebutuhan Pengguna Menggunakan Natural Language Processing (NLP) dan Multi Layer Perceptron (MLP)

Abstract

Selecting a laptop that aligns with the user's needs can be challenging due to the wide variety of available specifications and price ranges. This research aims to develop a laptop recommendation system based on user needs using Natural Language Processing (NLP) and deep learning approaches. The proposed system utilizes FastText for transforming user input text into vector representations and a Multi-Layer Perceptron (MLP) model to map user requirements into a corresponding vector representation of laptops. The dataset includes laptop specifications and a Suitable For column describing user categories. The data undergoes preprocessing and embedding processes to produce feature vectors. The model is trained using a cosine similarity loss function, and the recommendation is generated by computing cosine similarity between the predicted user vector and the dataset of laptop vectors. Evaluation is conducted using Mean Cosine Similarity. The results indicate that the system can generate relevant recommendations with an average Mean Cosine Similarity of 0.9396. The system also demonstrates flexibility by handling various forms of input, including mixed language, typos, and diverse user expressions, making it practical and robust for real-world use. Thus, the proposed system provides an effective solution for helping users choose a suitable laptop based on their daily needs.