News Recommendation in Real-Time - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 155

6.3.1.1 Visitors
Visitors represent the target group of news recommendations. They require
recommender systems to filter relevant items from large collection which they cannot
review themselves. Hence, recommender systems provide them value in terms of the
returned items' relevancy . Systems may determine how relevant visitors perceive
suggested news articles with different means. We may conduct surveys asking visi-
tors about the relevancy of their recommendations. This entails high costs. Therefore,
we may restrict surveys to rather small proportions of visitors. Alternatively, we may
evaluate visitors' dwelling times, return frequencies, or click rates.
6.3.1.2 Content Providers
We refer to content providers as editors in the context of news. Editors create and/or
select the contents to be distributed through news portals. They require recommender
systems to reasonably link news articles. Recommender systems ought not to confuse
readers with misleading suggestions but provide relevant resources. This reflects
the newspapers paradigm of structuring contents by grouping them in categories.
Visitors may expect to receive suggestion conforming to their previous interactions.
We may gauge recommendation algorithms quality in terms of representativeness
from editors' perspective. How well does a recommendation represent the previous
interactions? Alternatively, we may consider assessing how quickly visitors find
desired contents. For instance, wemay count howoften visitors immediately abandon
contents.
6.3.1.3 Advertisers
Advertisers strive to attract visitors. They want them to pay attention to their adver-
tisements and ideally buy their products or services. Typically, advertisers pay per
click. Although the click-through-rate fails to reflect their interests. Conversely,
advertisers prefer few clicks coinciding with a high conversion rate. Conversion
refers to visitors turning to customers. In our use case, we restrict our focus on click
rates. ORP does not provide access to data about visitors converting to customers.
6.3.1.4 Operators
News publishers pursue two main objectives. On the one hand, they try to distribute
informative and/or entertaining news to readers. On the other hand, they seek to max-
imize their rentability. This causes them to align the targets of users and advertisers.
Users have learned to ignore adverts on webpages [ 11 ]. Prompting users to continue
reading news increases the chances that they will notice adverts. Enlarged dwelling
times ought to lead to higher conversion rates. Consequently, news portals' earnings
will increase and improve their cost-effectiveness.