Collecting a Custom Database for Image Classification in Recommender Systems
Maria Vlahova, Milena Kirilova Lazarova · 2022
Nowadays machine learning and deep learning are widely used techniques for data analyses that require large amounts of labeled data. Moreover, with the migration from data as a service to data as a product the businesses are facing a complicated problem to collect the correct data from scratch that are suitable for data analyses and knowledge extraction using machine learning and deep learning-based approaches. The data collection and generation of custom databases is an active research topic aimed to overcome the major bottleneck for today’s business when machine learning and deep learning is to be applied for data analyses. In the paper a methodology for data collection is suggested that provide a structured approach for dataset collection allowing to overcome some of the challenges connected with the data collection activity. The suggested methodology is applied for collection of custom databases for solving an image classification problem in recommender systems.