Bibliographic databases/search engines

Question

Despite the -- or: due to the -- plethora of alternatives, I still haven't figured out which bibliographic database/search engine (Scopus, zbMATH, Google Scholar, ProQuest, EBSCO, JSTOR, etc.) offers the best coverage of OR/MS publications.

Also, assessing whether a paper presents what is considered the state of the art, cites highly relevant literature, a.s.o. are painful exercises.

So, my question is: What do you actually do when you browse the "interweb" for certain keywords, look up relevant publications in a specific (sub-)field, check for updated and/or improved results presented in a paper you've already read, etc.?

Do you use any of these databases/search engines listed above? If so, which -- and why, and how?

(Furthermore, if you are aware of a comparison/evaluation of databases/search engines from an OR/MS perspective, pointers are much appreciated.)

Answer 1

Ultimately, it depends on why you're doing the search. It also depends on what you mean by "coverage" - do you mean you want all papers that can be remotely considered OR/MS, or do you care about ensuring that you've captured everything from a specific (though potentially large) set of journals?

Google Scholar is fantastic for broad coverage, and is my usual port of call for when I want a a quick overview on how a specific technique is being used. However, as Dr. Trick pointed out this can be too comprehensive. Also, if you're concerned with paper quality and impact then Google Scholar has a tendency to add a lot of semi-relevant or poorly written papers to the results.

For these reasons, I'm a fan of the control that more traditional services such as Scopus give. I much prefer being able to restrict my search to specific journals or construct a boolean monstrosity to weed out a lot of these issues (on top of that, being able to easily export search results can be useful depending on why you're conducting the search in the first place). Of course, this isn't by any means comprehensive, which runs into the opposite problem.

No single database will ever be the "best", as they all have competing benefits and drawbacks. The best way to make sure you've got adequate coverage is to use multiple databases. Personally, my workflow tends to be something like:

1. Search on Google Scholar to identify relevant papers
2. Use the results of Step 1 to formulate a search string in Scopus (allowing restrictions to high impact journals, etc), adding these papers to the results identified through Google Scholar.
3. If need be, repeat from Step 1

This allows for a balance between including papers in other areas (such as linear programming in archaeology) whilst also ensuring that papers which can be trusted to be high quality as a result of their source are appropriately covered

Answer 2

For me, there is no substitute for Google Scholar. The main issue with our field is that it is pretty broad, so it is hard to tell where relevant publications might be. In many cases, the relevant literature might not be in an "operations research" journal. International Abstracts in Operations Research tried to handle that issue by collecting papers from a wide range of publications, identifying those that looked like "operations research". But this sort of hand-collection turned out to be difficult to do with the proliferation of journals and with the advent of automatic collection, so IAOR was shut down, though the site given has the last 25 or so years of collected abstracts.

Google Scholar has the following characteristics:

1. It is free. zbMATH and some of the others requires a subscription. My university does not subscribe to zbMATH and I can't see enough of it to determine whether it is worthwhile to try to push it with our librarian.
2. It covers both journals and conferences. A lot of operations research is now coming out in high quality, reviewed journals, particularly in those parts of OR that overlap with computer science. If you miss the conferences, then you miss much of the recent literature. Google Scholar does a great job of covering things.
3. It has great linkages. A Google Scholar search often leads to interesting finds by clicking on references of references of references.
4. The Author pages are really useful. Once I find someone in an area, being able (in many cases) to click directly on the author page gives a mildly curated view of what that person has worked on, with the ability to quickly find new (or old!) papers from that author.
5. It covers everything. If there is a paper in the archaeology literature that is relevant to OR, it will be somewhere in Google Scholar. As I write this, I just got lost for 10 minutes in Google Scholar looking at the archaeology literature, including papers like this.

A couple disadvantages:

1. It is a little too comprehensive. It can be hard to distinguish quality publications from non-quality publications. Some questions on OR.SE come from trying to understand papers that are essentially nonsense. But the paper is in Google Scholar, and other than citation count there is no mechanism for distinguishing it from a paper in a high quality journal. This is particularly important when looking outside my own area for relevant references.
2. As a free resource, Google Scholar can go away overnight if Google decides that is advantageous for them. Having been burned with Picasa and Google Reader, I am cautious about putting all my eggs in Google's basket.

Answer 3

I agree with Dr. Trick that I tend to start with Google Scholar. If I'm looking to see what's been done since a paper was published, I'll often click the "cited by" option.

Another option that may be helpful is the preprint site optimization-online.org. Sometimes I'll do a quick search there too.