With Google Scholar, PubMed, and other free academic databases at their fingertips, scientists may feel they have plenty of resources to trawl through the ever-growing science literature.
But a search engine unveiled on 2 November by the non-profit Allen Institute for Artificial Intelligence (AI2) in Seattle, Washington, is working towards providing something different for its users: an understanding of a paper's content. "We're trying to get deep into the papers and be fast and clean and usable," says Oren Etzioni, chief executive officer of AI2.
The free product, called Semantic Scholar, is currently limited to searching about 3 million open-access papers in computer science. But the AI2 team aims to broaden that to other fields within a year, Etzioni says. His team is well financed: AI2 was founded and is backed by Microsoft co-founder Paul Allen, who has given the institute more than US$20 million since 2013.
Semantic Scholar offers a few innovative features, including picking out the most important keywords and phrases from the text without relying on an author or publisher to key them in.
(Score: 1, Interesting) by Anonymous Coward on Tuesday November 03 2015, @10:26PM
They should create graphs of who has published with who on the results returned from the search. Node thickness could be total publications for each author, edge thickness could be some function of number of shared publications (take the log +1 or something). Split it into communities too. You can even make videos of the publication network growing over time (similar to this: https://www.youtube.com/watch?v=dTILX-_JzTs). [youtube.com] If it can be interactive, have a pop up on hover showing most common terms, etc.
I did this while doing background research on a neuroscience topic before, and it was very informative. Especially to identify and find the papers from the outgroups. These were much more skeptical than those from the main players, who are also easily identified. One problem will be identifying unique authors with Chinese names, many duplicates there requiring manual tinkering. Also, some papers are going to be irrelevant so there needs to be a way to exclude them. Such a tool would be very useful for literature reviews, I never got it automated enough though.
(Score: 0) by Anonymous Coward on Wednesday November 04 2015, @04:35AM
They should create graphs of who has published with whom on the results returned from the search.
The more you know!