acm-header
Sign In

Communications of the ACM

ACM News

Google Scholar Pioneer on Search Engine’s Future


Google Scholar creator Anurag Acharya.

Google Scholar co-creator Anurag Acharya recalls the service's history, and discusses his outlook for its future.

Credit: Amit Basu

Google Scholar, the free search engine for scholarly literature, turns ten years old on 18 November. By 'crawling' over the text of millions of academic papers, including those behind publishers' paywalls, it has transformed the way that researchers consult the literature online. In a Nature survey this year, some 60% of scientists said that they use the service regularly. Nature spoke with Anurag Acharya, who co-created the service and still runs it, about Google Scholar's history and what he sees for its future.

How do you know what literature to index?

'Scholarly' is what everybody else in the scholarly field considers scholarly. It sounds like a recursive definition but it does settle down. We crawl the whole web, and for a new blog, for example, you see what the connections are to the rest of scholarship that you already know about. If many people cite it, or if it cites many people, it is probably scholarly. There is no one magic formula: you bring evidence to bear from many features.

Where did the idea for Google Scholar come from?

I came to Google in 2000, as a year off from my academic job at the University of California, Santa Barbara. It was pretty clear that I was unlikely to have a larger impact [in academia] than at Google — making it possible for people everywhere to be able to find information. So I gave up on academia and ran Google’s web-indexing team for four years. It was a very hectic time, and basically, I burnt out.

Alex Verstak [Acharya’s colleague on the web-indexing team] and I decided to take a six-month sabbatical to try to make finding scholarly articles easier and faster. The idea wasn’t to produce Google Scholar, it was to improve our ranking of scholarly documents in web search. But the problem with trying to do that is figuring out the intent of the searcher. Do they want scholarly results or are they a layperson? We said, "Suppose you didn’t have to solve that hard a problem; suppose you knew the searcher had a scholarly intent." We built an internal prototype, and people said: "Hey, this is good by itself. You don’t have to solve another problem — let’s go!" Then Scholar clearly seemed to be very useful and very important, so I ended up staying with it.

 

From Nature
View Full Article

 


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account