Some aspects of the use of recommendation systems

Recommender systems daily influence decisions made in the process of using Internet resources. While the development and implementation of significant attention is paid to issues such as accuracy and privacy of recommendations, the long-term mutual feedback between recommender systems and user recommendations has never been subjected to serious research.

Despite the widespread belief that recommendations help users find something new, the continued use of recommender systems can contribute to an extraordinary increase in the popularity of certain elements and, ultimately, narrow the user's choice. These results are supported by some studies in real systems.

Even if we do not notice the effect, our online life depends on the recommendations. Popular websites like Netflix, YouTube, Amazon and Ozon, in an effort to facilitate navigation, offer the possibility of providing relevant elements. Thus, the growth of user satisfaction is achieved and, most importantly, profit. Today, various working algorithms of recommendations are widely distributed, ranging from the simplest variants “buyers who chose X also acquired Y” to complex variants like singular decomposition .
')
Although many users still operate independently of any automated assistance, the use of recommender systems is steadily increasing every day. The main feature of any recommendation system is the ability to match customer requests with relevant products. This task is especially important and difficult for less popular elements for which user patterns cannot be easily identified. Correct matching of less popular elements is crucial for e-commerce. Studies have shown that from 20% to 40% of sales on Amazon are not among the most popular products. If you rank the goods and compare their sales, you can get a certain schedule with a long tail of distribution, which contains a large number of niche elements. Often they provide a higher level of profit compared to the most popular products.

In this regard, recommender algorithms make it possible to reveal hidden distribution resources with a long tail . Recommendations work by expanding the variety of recommended items and more evenly spreading the user's attention. However, at the moment, the algorithms implemented on many popular resources do not fully perform their functions because of which the tail of the popularity distribution becomes shorter. At the same time, the most popular items still account for a significant share of total sales. Such an adverse effect, ultimately, can lead to loss of balance in the system.

Algorithms based on precision-oriented metrics cannot, by their nature, explain this behavior. Although, if you look more closely, you can find feedback between the choice of users and the system of recommendations. This interaction is similar for many physical systems. Therefore, for research it is quite acceptable to use a physical approach.

Currently, recommender systems are most often investigated for achieving short-term indicators such as accuracy and diversity, and it is quite obvious that the use of conventional recommender algorithms results in the system reaching a stationary state, in which users focus on a small number of elements (goods), rather than distributed over a wide range. In other words, the usual recommendation algorithms ultimately narrow the user's choice and reduce the information horizons instead of their expansion.

In some cases, there is also a hysteresis phenomenon, which implies a serious dependence of the system on the initial state. At the same time, this indicates the insufficiency of the current recommendation systems for online stores, applications, search engines, social networks and the media. And also about the necessary compromise between the short-term and long-term effects of the use of recommendation systems in the design and implementation of the next generation of recommender systems. So at the moment it is necessary to engage in more extensive research in this area, since the gain from a more complete use of the entire distribution can be huge and provide a kind of reserve for the entire online business.