Abstract:

Long-tail data-based recommender systems are gaining attention for their many applications in online services. Such systems demonstrate their ability to more accurately take into account the interests of network users and open up great prospects for better meeting their needs. This article discusses some approaches to using the information found in the long tails of web users' requests.

This article answers the question, why are the tail queries of users starting to play an important role in recommender systems? And to what extent they correspond to the interests of netizens and how they are taken into account and participate in the development of recommendations.

The hypothesis refers to long-tailed data (LTD) as a way to improve the accuracy of recommendations by using the LTD by the discriminator system (DM) to validate the recommendations (R) proposed by the generative model (GM): LTD®DM®GM®R. The long-tail data concept improves the reliability of recommendations, expands the user base, improves service availability, and improves the accuracy of user behavior assessments..