With over a million new publication every year, it can be hard to keep up. For example, one of the keyword I use for new publication alert is “mucus”. This search query can hit over a hundred new paper each week on pubmed!
So I have to go through every single hit, do a rapid scan of the title and judge if the whole paper is relevant to me or not. This is clearly un-optimized was to stay updated with the latest papers: it is time consuming, and who knows what may be hidden behind a seemingly uninteresting title. You can try to optimize of the search queries, but you then narrow down your result and still risk to miss out on important papers.
Now what if smart algorithm could understand what you are looking for, based on your publication and who you are working with? Pushpin is the result of a graduate work by Dr. Wolfgang Reinhardt. It includes a classic social network interface, with profiles, publication list and so on. But Pushpin also makes recommendations regarding who you should follow and what papers might be interesting to you.
Pushpin does these recommendations by carefully analyzing what your field of interest is. This is done based on keywords and publication rating you can enter in Pushpin, but also on the similarities between the full texts of your publications and the work of others.
This service has the potential to be a time-saver for researchers, filtering out irrelevant literature. A similar service was rolled out by google scholar last year. I have not tested the Pushpin service to it full extent, but it would be interesting to compare it with its google counterpart.