| |
At the present stage of social development, recommendation systems are an integral element providing interaction between the user and the producer of information. Such systems are closely connected with public life and are used for various purposes, including e-commerce, social networks, and mobile phone applications. In the context of the global development of artificial intelligence based on deep learning of neural networks, there is an objective need to analyze the possibilities of using neural networks in recommendation systems.
The author concludes that if compared with traditional models of training recommendation systems, models based on neural networks and deep learning provide great opportunities for recommendation systems. This is because neural networks, due to the possibilities of in-depth analysis of various disparate data, as well as their ability to learn from their mistakes and improve in the learning process, allow them to obtain very accurate data. Hence, recommendation systems can be more adapted to the needs of a particular user and will allow you to select preferences even in those goods, works, services that the user has not previously ordered or was not interested in. However, the main problem in this area is the correct choice of a neural network and training technology, as well as the selection of a data set on which the training of the corresponding neural network will be carried out. The spontaneous choice of a neural network for a recommendation system, without taking into account the above parameters, can lead to errors in the analysis of data by a neural network and its long-distance learning, and as a result lead to incorrect recommendations for users.
Keywords:neural networks, recommendation systems, user preferences, information filtering, data analysis, deep learning
|
LEGAL
INFORMATION:
