You’re staying up to date with the current literature in your field: good. You might use journal RSS feeds or get journals’ table-of-contents (TOC) mailed to you. If you’re lucky, a friend will have a lookout on new interesting papers.

Unfortunately, these methods have a limited scope as they don’t cover the papers of authors or journals that you’re not familiar with. The following services enable you to get in touch with new material in a new way, namely personalised recommendations.

In essence, personalised recommendations work by knowing what your research of interest is, by either having access to your personal library or saved search queries. They work in two ways: the first is library based (first two examples) and the second is search-query based (following three examples):

Google Scholar: My Library

Set up a Scholar profile and specify your publications and Scholar will build a library for you called ‘My Library’. This library will contain your own work as well as the papers that you have cited. You can add papers yourself within a Scholar search query (hit the ‘Save’ just below a result). Based on these contents, you will receive a notification of interesting papers by mail or in the search interface. Despite lacking the option of uploading your entire library, this is still the best service as the recommendations are spot-on.

PubChase

Register at PubChase and sync your Mendeley library. PubChase will read your library and make recommendations based on its contents. It’s much easier than Scholar, as the synching of your library happens automatically. The recommendations are nearly as good, and their support is incredible. When you are already using Mendeley, give this a shot!

There’s also ReadCube, but I haven’t tried that one (yet), as the functionality is the same as Mendeley+PubChase (old comparison)

Sciencescape

Register at Sciencescape and indicate your topics of interest. Sciencescape is the only service of the bunch that adds a historical sense to a provided list of recommendations. It looks handsome and has a smooth interface, which is quite special compared to the rivals. What is more unique, is the weight it adds to the recommendations. Based on the historic site of the author’s lab and journal’s reputation, it will apply weights. The major downside of this service is that you can’t upload your library or customise search queries. Furthermore, it takes time to build a list of recommendation and the basis of recommendations focuses on the more molecular sciences.

Sparrho

Register here and make search queries. Based on your search queries, Sparrho will provide you with new literature. I fiddled around with it for a bit but didn’t find any useful papers soon (unfortunately, I think, because the lay out is sweet). You can set up different search queries – but it will only give you suggestions per query and not combined. So make sure that your search query is already focussed on the subject of your interest!

Scizzle

Register and specify your (combined) search queries, and Scizzle will search for papers and suggest some. You can add the papers to your ‘scizzling pot’, which is an extra source for new recommendations. I think it makes the hassle of building new queries quite funny, but the interface is slow and you have to do a lot of labour yourself.

And there’s Nowomics but I’m not working with genes. And there’s CiteULike (I can’t get this thing to work, even after uploading my entire library).

Concluding, my recommendation:

It actually depends on your circumstance on what I’d recommend: if you’re starting a project, without having read any papers, working query based will get you quite far. But, if you have already have done a couple of projects, do stick with the library based ones, as this is a more detailed reflection of your interests. I’ve used GS:ML and PC for quite a while and am really happy with what they come up with. The other services came to me via a Nature paper and have tried them in the past weeks, but I’ll not likely continue with them as I find the interface quite time consuming.

Next Post