In:  Guest Blog  

Drowning in information? Throw me a recommendation life ring!

Image credit: Peace by OC Always, Flickr. CC BY
Image credit: Peace by OC Always, Flickr. CC BY

Here at ScienceOpen, we like to showcase new technologies that may improve the efficiency of research, especially those that focus on speeding up access to information, preferably of the open kind!

We were alerted to the scientific recommendation engine called Sparrho, by a tweet from John Wilbanks (great admiration for this chap). We invited Qingzhi Fan, who has a PhD in Biochemistry from the University of Cambridge and has worked in finance and as a consultant, to explain what Sparrho does and why it of benefit to researchers. Now over to Qingzhi…

Scientists, this is good news and bad news. Whether we like it or not, scientific information has been transformed by the digital age – this means quicker traffic to a greater volume of content than ever before. The challenge of sharing ideas with a broader audience used to be its dissemination; but today, it’s how to grab the right attention, in the right way.

From a scientist’s perspective and as a scientist myself, staying up-to-date with the latest research in my field is as important as working hard in the lab. The wealth of information available today helps speed up breakthroughs, but at the same time confuses, distracts and overwhelms us. There are two issues here: where to find the relevant information and how to do it quickly.

As Richard Van Noorden pointed out in Nature, scientists may be reaching a peak in reading habits. We are adapting: moving away from library and traditional paper prints to read online, moving away from verbose articles to prefer short succinct ones, moving away from reading articles in full and in detail to power-browsing. Yet, reading time itself is often not the real frustration, but the time wasted before finding any relevant information.

The recent emergence of various recommendation services that help researchers stem the rising tide of literature is well described by Elizabeth Gibney in Nature.  The idea of using these tools is to be presented with the relevant information without having to look for it, then our job becomes to read and interpret it, even to share it with others.

For example, our scientific discovery platform aggregates and distills information based on user preferences and makes personalised suggestions. The algorithms are designed to learn user needs and go beyond linear keyword search in what can be described as a three-pronged approach:

1) Data-data analysis: using techniques like natural language processing to pull the most relevant research based on the data users provide (keywords, subject area analysis, etc).

2) User-data analysis: every time users interact with the site by indicating what information is ‘relevant’ or ‘irrelevant’ to them, the engine can build up an article-to-field level of understanding in terms of users’ interests.

3) User-user interactions: users act as the intelligent curators of the recommendations; technology is merely the enabler. Individual user interactions not only improve their own recommendation profile, but also help the whole user community with similar interests.

A good recommendation tool can go beyond scientific articles and act as a one-stop shop for researchers. It will recommend relevant talks, seminars, conferences, posters, patents, grants etc and aggregate all categories of the latest news with filtering functionality. Users can set up automatic newsfeed and not worry about searching and missing the latest information; at the same time, these services often provides opportunity for serendipitous discoveries hidden in places users never normally look.

To every problem there is a solution. For content recommendation, the solution may not be initially perfect since it only gets better with more user interaction, but it may be the life ring you need to stay afloat.