Since 2005, Simon Inger Consulting has periodically captured a snapshot of the attitudes and opinions of users of the scholarly literature via a standardized survey.

The fourth iteration of the survey was completed in 2015 and the results (published in March 2016) make interesting reading, with certain long-term trends now becoming apparent.

Specifically, the authors of the survey note that: “Related Articles functionality is the only area that has become more useful [for content discovery] since 2012”

That is, while the proportion of users reporting Search and eTOCs to be useful functionality for a journal website has decreased, the number reporting Related Articles as important has increased. In fact, looking at the 2015 data, for the first time Related Articles is ranked most highly of all the journal web site features listed.

This should not be a surprise. Sifting to find relevant information in an ever-expanding online sea of information is a major challenge. The most successful consumer platforms, including Netflix, Spotify and Amazon, have invested heavily in algorithmic recommendation technology to guide users to content of interest.

Scholarly publishing has recognized the trend. Bibliographic services and tools such as F1000 and Mendeley emphasize content recommendations, and newer startups such as Meta and Sparrho are focusing specifically on providing researchers with highly focused reading lists, tuned to their needs and interests. Recommendation technology continues to be an active area of theoretical and applied research.

TrendMD’s “article recommendation widget” is designed to make it easy for any scholarly publisher to deliver high-quality recommendations, just by dropping a few lines of HTML into their article template. By using collaborative filtering technology, TrendMD is able to take advantage of patterns in click data to identify the most interesting and relevant articles, even if they are not necessarily the most “similar” in semantic terms, leading to dramatically higher click through rates on recommendations compared to pure semantic solutions, with commonly 4-5% of visitors to a page clicking on one of the recommendations shown in the TrendMD widget.

No publisher is an island, and the world of research is an interconnected one. This is the key to the second key benefit of TrendMD’s recommendation technology, which is the ability to include both ‘internal’ recommendations (to a publisher’s own content), and external recommendations (which link to and from highly related articles on other publishers sites). For the user, this turns the scholarly literature into a more connected and explorable universe of information, while for the publisher it opens up an important channel to reach new readers.

As the study reported above makes clear, article recommendations are more important than ever for users of the scholarly literature, and deserve increased attention from publishers. TrendMD’s best-of-breed solution allows any publisher to make the most of this trend by delivering state-of-the-art content recommendations.