Juan Carlos Perez (IDG News Service) recently spoke with David Lipman, director of the National Center for Biotechnology Information, part of the National Library of Medicine, which hosts various life sciences databases, including the PubMed scientific literature database. Lipman discusses the emergence of Google and Yahoo as scientific research tools, as well as major improvements on tap to PubMed’s search functionality.
Some people view Google, etc., as increasingly useful research tools in the life sciences, joining PubMed and other established services. What is your opinion?
Lipman: People sometimes have this a bit backwards. The primary point is that there is a huge amount of information on the Web. The best way to make the Web more useful for research and education is to get more information out there. What we find on our site is that the more good-quality content that we add to a resource, the more it is being used. If you have more gene sequencing data, you have more searching. If you have more full text in PubMed Central, you have more searching.
While I want to give MSN and Yahoo and Google a lot of credit for doing a good job, the reality is that if there wasn’t content there, it wouldn’t matter how good those search engines were. And if the search engines were a little less good than they were, you would still be getting to a lot of good content... And if those search engines got better, it wouldn’t make as much of a difference as if there was more good peer-reviewed medical content. That would make a bigger impact.
What do you think about Google Scholar?
There is a lot of involvement of the library community in PubMed so it’s more transparent in terms of what could be there and isn’t there… So when you don’t find something, it might mean something because [PubMed] is updated [often] and its scope is well defined. When you search Google, Yahoo, or Google Scholar, it’s more open-ended what could be in there, and it’s not clear – when something isn’t there – why it’s not there. These things aren’t good or bad per se. It just means that there are advantages [in both types of services.] I’m thrilled when I see Google make the effort to do something with Google Scholar because I know it’s going to mean more people will make more discoveries. A lot of times they’re sending people to us as well, and that’s great.
Are you planning usability enhancements to PubMed?
We do a lot of things in PubMed to make it so you don’t have to know much and you can do a query and we’ll do a lot of interpretation of what you’re doing to try to make it simple. Although the interface doesn’t look that simple, if you just type in some stuff, we’re doing a lot of backend work to map things to other terms to make it easier. The problem we face with PubMed and our whole system is that to gain more of the advantages, you need to know more things. You need to poke around more.
We’re definitely looking at our user patterns more carefully now. You’ll see over the next year some significant changes in it, not so much to make it easier to use but to make it easier to make more discoveries with it. If you type something in, you’ll see more suggested additional links to look at … We can connect you up to related information much more effectively than we’ve been doing. We’ll make it easier to use in very powerful ways. I want more people to make more discoveries.
Are you talking about identifying search patterns from users so you can surface links they might be interested in?
If you did a search on, say, Alzheimer's disease, and you find a PubMed record that looks interesting to you … right now you go to that page and we may have done an incredible amount of work to link that up to all kinds of things – to information about genes that might be involved, to genetic versions of it, to related articles. And the reality is you have to poke around to see that – it’s sort of hidden in a way.
What we’ll do instead is, using computational tools, poke around for you and then put them right on the record. A lot of this is based on a much more careful analysis of how people are really using the site and what they’re really doing, and [questioning] why they aren’t doing some things more than they are. It’s about making the computer do a little more work so that the user just sees it. It’s just there and they don’t even have to think about it.
The bottom line for us is discovery. We want people to make discoveries, and if we’re using up real estate on the Web page for things people don’t click on, and if we can put things on there that would have been associated information, then we should do that.
Is there a specific initiative tied to these improvements?
Yes. I call it the Discovery Initiative. It’s something new. It’s been percolating. Last summer, I went out to visit Google and Amazon’s A9 [search engine] and the folks from Microsoft’s MSN came to visit. I also went up to Boston to meet with folks from the major hospitals there and MIT and Harvard. We’ve really been giving this a lot of thought. In many ways, we have had great success. Lots of people use the site and 2.25 terabytes of data are downloaded from our site everyday. And yet I find it very frustrating because we’ve connected up the scientific information in very precise and powerful ways: a protein structure to the chemical it’s bound to, to genetic data, you name it. All that is connected up. And yet very few of our users do more than very simple things with our site. It’s as if [they say], “That’s enough, I found the answer I’m looking for and I’m done.”
We want them to find answers to questions they didn’t even know they had. They can do that. People who are really experts at using this site can make discoveries almost on demand. … So this is a new initiative. We’ve always been working on making incremental improvements, but now I’m talking about something that is a more powerful shift. I’m hoping that without the users having to think about having to learn anything different, they’ll be able to take advantage of databases they didn’t know about and [make] connections between things they don’t have to ask for anymore. This is a new phase.
What stage is the initiative in right now?
We’ve been re-engineering our backend to make it more flexible on the front end and user interfaces and that’s moving along. We’re almost ready to roll that out. That wouldn’t necessarily be seen as a difference, initially, by users, but it does give us more flexibility in terms of the front-end stuff. A number of these issues we had [identified] as things we wanted to work towards over a year ago. Some of this is still in the discussion phase, some is very far along in terms of implementing, and some inevitably is going to be an experimental aspect. I’m hoping within a few months some of the initial beta aspects of it will be out there.
You said most changes will be implemented this year, but is this a multi-year effort?
It marks a change in perspective and philosophy that will lead to constant changes in the system [in coming years].