***************************************************************** * * * The Literature Search 'Till the End of the Universe! * * * * The 2nd guide in the Hitchhiker's trilogy * * * * Ron Azuma Don't * * v. 1.01 Panic! * * June 1993 * ***************************************************************** ================================================================= "There are no answers, only cross references." - Weiner's Law of Libraries "Get thee to a library!" - John McHugh, telling me how to start preparing a dissertation proposal ================================================================= => Who should read this document? This guide is intended to help students who are doing literature searches. It distills some of the things I've learned while doing literature searches for my own Ph.D. proposal and other documents. The literature search is an integral part of research. Without it, you can't determine if your "novel" method is really new, what other scholars have already tried before, or learn who the other players in the game are. A literature search is a prerequisite to publication and dissertation work. => But I already know how to use a card catalog. This guide does not presume to lecture you on the basic stuff. Instead, it describes resources you might not be aware of, provides tips so you can avoid mistakes I made, and discusses more pragmatic issues (like how to visit the libraries at Duke and State without getting parking tickets). => Ok, sounds good. Where do we start? I'll assume you have a research topic you want to investigate, that your advisor has already given you some references to start with, and that you can find all the easy ones in Brauer or Davis. I'll also assume you've found additional references by backchaining from this initial cache of papers and acquired the easy-to-obtain papers from that list as well. Gee, that was easy. Now the fun begins. Now you have a pile of references that you can't find on campus. Maybe they are technical reports, or are from a conference proceedings that we don't carry, or it looks like a report printed by the government. Also, you're worried that your list of references isn't complete, that there are other papers out that are relevant but you don't know what they are. How are you going to find them? What follows is a loosely-organized list of resources and tips that can help you achieve these goals. Also, a few disclaimers. I'm no expert in this field; I've just had some experiences in tracking down some hard-to-find sources for my own studies. This is no substitute for advice from your friendly neighborhood research librarian. As with all Hitchhiker guides, this is entirely unofficial and not sanctioned by the Department in any way. I do not guarantee that the information here is correct and you assume all risk for following any advice within this guide. You can thank the 50 million lawyers in the USA for this paragraph. ========================= Finding additional papers ========================= If you don't have a list of references to start with, or you want to find additional papers that aren't listed in your references, there are several indices that can help you: * Index to Scientific & Technical Proceedings [ISTP] There are hundreds of conferences held every year, many of them obscure. This index, sorted by year, lists most of the conferences that were held in that year and includes a list of all of the papers that were presented in each conference! The conferences are cross-indexed under subject and author headings, so you can look for all the conferences that cover a given subject. You can find this in Davis, in the annex above the reference section. * Applied Science & Technology Index [AS&T] Also in Davis reference annex, this index covers scientific and technical papers in popular magazines and journals (e.g. Time, Scientific American, etc.) While this isn't the first place to look for papers, it can provide useful information, especially if you're looking for an introductory article. It looks and works much like the Reader's Guide to Periodical Literature, which you probably used in high school. * Engineering Index * ACM Guide to Computing Literature * Index to IEEE Publications These three indices list papers from many different academic journals, and the three combined will do a decent job of covering almost all CS journal publications (since most are sanctioned either by ACM or IEEE). You can look up papers by subjects or CR categories (in the case of ACM) and they will tell you what journal they are from. The Engineering Index is found in the Health Services Library, and the other two are in Brauer. * Computer & Control Abstracts Another index that references CS papers. I haven't used this much. It's in Brauer. * Computing Reviews Recent issues seem to give critical reviews of recently published CS books. It's in Brauer. * Government Reports Announcements The US Government sponsors a huge amount of research (through grants to industry, govt. labs like Lawrence Berkeley or NASA, and universities). If it's funded by the government and is not classified, then the reports from those projects must be made available to the public. How do you find out about these documents? Through Government Reports Announcements, where the National Technical Information Service [NTIS] lists all of the documents that are available. It's in the far corner of Davis' reference area (or you can find more on-the-shelf copies at Duke). Here you can find a wealth of project summaries, papers, and other technical documents, and each of them has an abstract and a page count! At Duke Engineering, there's a similar set of announcements for just NASA documents; I don't remember the name of it though. ========== Microfiche ========== So now you've found some interesting government documents and you want to get hold of them. You can order them directly from NTIS, but each paper will cost you between $10 and $60. That will break anybody's budget real fast. There is a better way. The government maintains sets of these documents in repositories across the country. They are stored on microfiche. Theoretically, Davis library is one of these repositories. They certainly advertise it well enough, with the sticker blazing prominently on a wall. But in reality, Davis stocks virtually nothing. The real place you want to go to is D.H. Hill library, at State. Each paper that you find will have an associated ID number. Examples of what they look like are: AD-A222 119 PB88-140321 N88-17600 Go to the govt. documents section at D.H. Hill, which is somewhere on the 2nd floor. A librarian can point you to a card catalog that lists all of the documents they have. Look up the ID number of each paper in the card catalog to see if it is in stock. If it is, you can fill out a card and give that to another librarian, who will fetch the microfiche that hold your paper. You can look at them on nearby microfiche readers, and then print the pages on a thermal printer at normal photocopier rates (around 7 cents per page). So for an evening's effort, you can print hardcopies of all your papers at very reasonable prices while also being able to look at the paper first before printing it to verify that it's really something you need. ================= Technical reports ================= Technical reports tend to be hell to find. They're not published, so libraries rarely carry them. Natalie Sipes is the person to ask in this Department for help in locating tech reports. Look in the /usr/doc/local/techreports directory for information. In there, you'll find: * Lists of tech reports from other universities and companies * Contact information for other sites, with addresses and sometimes phone numbers and E-mail addresses * A list of all the reports produced here. Many of them are in the 2nd floor reading room. Some are available on-line. There's also a company called Research Access, Inc. that distributes some technical reports. I've never contacted them, but they are listed in the directory: Research Access, Inc. 3400 Forbes Avenue, Suite 302 Pittsburgh, Pennsylvania 15213 1-800-685-6510 urr@andrew.cmu.edu More often than not you'll have to contact the source institution if you want one of its tech reports. ====== Theses ====== You can look up abstracts of the CS dissertations produced each year; I believe these are in the ACM Guide to Computing Literature. A few highly noted dissertations get published by places like the MIT Press. But most are just printed up as technical reports by their respective CS departments and can be ordered that way. But note that not every department prints dissertations as technical reports. Some departments also publish dissertations through University Microfilms. If they have the dissertation you are interested in, they will photocopy it and mail it to you, for a fee. Contact them at: 300 North Zeeb Road Ann Arbor, MI 48106 (800) 521-0600 You might want to ask your reference librarian first about how to order stuff from University Microfilms, since I've never done it. Interlibrary loan is another option for obtaining theses. If it's a thesis produced at UNC, then one of the libraries on campus must have a copy. All CS theses are maintained on the upper level of Brauer. ============================== Libraries at Duke and NC State ============================== If you can't find certain journals, books, or conference proceedings here at UNC, you might be able to find it at Duke or NC State. Even if you root with all your might against both schools at football or basketball games, you will probably have to visit one or both to hunt down missing references. Why? Because both Duke and State have an engineering program, and we don't. But don't feel guilty about sneaking onto their campuses and using their resources; it works both ways. You can be sure that Civil War scholars at Duke and State flock to Wilson library, for example. We're actually quite fortunate to have three such library systems so near to each other. Individually, each has weaknesses, but combined they are an impressive resource. You can conduct an electronic card catalog search at Duke and State simply by adding "all" before each search. You can also check the standard hours for all the libraries by typing "help locations" and selecting the library you're interested in. For additional information and finding special hours during holidays and when classes aren't in session, they provide phone numbers for you to call. The two main libraries of interest at Duke are Perkins, the main general library, and Teer Engineering library. At State, they basically don't believe in having lots of separate libraries, preferring instead to cram everything into one megalibrary: D.H. Hill. We have interlibrary loan privileges with both Duke and State. You can get an interlibrary loan card from Davis by presenting your student ID. At Duke, I believe you need it validated at Perkins before you can borrow anything at Teer. Where are these libraries? You can get campus maps of Duke and State by going to Davis and borrowing the course catalogs for the two schools from one of the reference desks. Campus maps are inside the catalogs. Teer Engineering is on Science Drive, and Perkins is close by, next to the Chapel. D.H. Hill is a huge building right on Hillsborough Street. We have many alumni in the Department who are from these two schools, so track one of them down and ask for directions. They may keep a low profile during ACC Tournament time, but you can generally find one. :-) Where and when can you park? Unfortunately, parking at Duke and State is limited, much in the way parking here at UNC is. Like UNC, parking lots seem to open up around 4:30 or 5pm and are usually open on weekends (if no sporting event is taking place). Lots can fill up in the evening as well, because of evening classes and students coming in to study. Generally, I try to time it so I arrive around 5pm when the lots empty out and before the evening crunch arrives. At Duke, if you drive on Science Drive toward Teer, the parking lot on the right immediately after you pass Teer is very convenient for visiting Teer and Perkins. If that is full, there's also space behind Teer. Visiting around 4:30 or 5pm is important for another reason: the doors to the outside of Teer lock around that time, and then you have to knock on the door or wait for someone to let you in. At State, I drive down Hillsborough, take a left on Brooks Avenue and park in the lot on the immediate right. I don't know if that actually opens at 5pm, but I've never been ticketed and it's usually empty at that time. If that is not available, you can cruise the neighborhood streets within a few blocks north of campus. Read signs to see when parking is allowed. Also, D.H. Hill itself has a tiny lot next to Hillsborough, but I've never had much luck finding a space there. Once I parked in a lot on the far east side of campus, which looked like a former stadium. Show the guard your ID and say you're visiting the library, and he may give you a visitor's permit. If you haven't visited Perkins, Teer, or D.H. Hill, give them a try! ================= Interlibrary loan ================= I've had some success with interlibrary loan for getting theses, and you can also request books or other materials. If you can't find it within the UNC system, you can go to the reference desk at Davis and fill out an interlibrary loan request. It can take several weeks before your material arrives, though, so don't do this if you need it for a project due in three days. If it's at Duke or State, don't use interlibrary loan unless you're really lazy. Just go there yourself and pick it up. ========= Companies ========= Most companies have a corporate library for their own employees. These can be excellent sources for certain specialized types of information. For example, when I worked at Apple I noticed they had every manual ever printed for an Apple computer, and most of the advertisements ever shown on television. I've been in the Pentagon's library, and you can guess what was there. For a local company, you can contact the head librarian, politely explain what you are looking for and why you need it, and then ask to visit. You probably won't be allowed to check out anything, but there's a decent chance they'll let you in. If not, or if the library is far away, then you'll need a contact to do the leg work for you. For example, I wanted to get papers out of some flight simulation conferences that weren't in any library that I knew of. So I spoke with contacts at a few flight simulation companies and they were able to photocopy and send me some of those papers. I probably would not have found them any other way except paying $10 per paper (ouch!) to the conference organizations themselves. If you do visit a corporate library, try to work on your own as much as possible and avoid bothering the staff, since they get paid to aid the company's employees, not you. You can contact the Special Libraries Association at (202) 234-4700 for help in identifying special libraries. =============== Human Resources =============== An hour with an expert can be worth 20 hours in the library, especially when you're starting out. If you need to explore a field outside CS, don't be afraid to contact a professor in another department on campus and ask for a brief meeting. Most will be quite happy to talk with you. Don't ignore experts outside of campus, if relevant. E-mail is a wonderful tool for reaching some of them. Getting in touch with off-campus experts tends to be more problematic, however. My experience has been that many ignore me or brush me off. Sometimes having your advisor make the first contact increases the chance of actually speaking to the expert... ============= Computer help ============= Maybe one day all journals, books, and conference proceedings will be digitized and stored on computer, so you can do a complete search and retrieval without leaving your office. Maybe it will even be easy to do. Maybe the Chicago Cubs will win a World Series. What's possible today? Certain indices are stored on CD-ROM and are accessible via dedicated PC's at Davis. Duke and State also have CD-ROM services available. These can be faster than checking indices the old-fashioned way. Of course, you can access the electronic card catalog without leaving your workstation by doing: telnet librot1.lib.unc.edu More sophisticated computer searches are available, where you specify some parameters and then have the computer search multiple databases and return a list of papers to you. I have not tried this, but know people who have. It has the potential of saving you lots of time, but it can be expensive (over $100), especially if you go through private services. They can charge you by time or by the number of paper titles that you retrieve. You can access some on-line databases through your home computer and a modem. I really don't know a lot about this area, so you'll have to ask your librarian for more information. Supposedly one can ask reference questions to Davis Library personnel by emailing to question.davis@mhs.unc.edu ...although I haven't tried this. ==== Misc ==== Some of the material I've mentioned comes from a book called _Find It Fast_, by Robert I. Berkman (Harper & Row, ISBN 0-06-096486-3). It's aimed more at a reporter looking for general information than a doctoral student searching for previous academic works, but it's a useful handbook nonetheless. The online catalog at UNC, Duke, and NC State is not the easiest to do searches on. The one from my undergrad, Berkeley, is much easier. You can access this catalog by doing: telnet melvyl.berkeley.edu or telnet melvyl.ucop.edu Using a terminal code of vt100 seems to work for me. Why is Cal's online catalog (called Melvyl) better? If you search on title words "redundant file systems" in the local catalog, you will get back only the titles that START with those three words, in that order. But if you search on Melvyl, it will return all the titles that have those three words in any order and in any position in the title! Another way to get at online catalogs from other universities is to use laUNChpad. Just telnet to launchpad.unc.edu and login as launch. After choosing LIBTEL from the menu, you can navigate your way through many online catalogs from North American universities. Literature searches take time! They are tedious, labor-intensive, and usually not much fun, but they are a vital component of research. This guide is by no means complete. If you learn of other sources or have different tips, send them to me so I can include them in future revisions of this guide. => Thanks for all the tips! But how do I know when I'm done? You don't. It's like the Halting Problem; you can never be sure that you've found all of the relevant papers. The Dec. 92 issue of Communications of the ACM estimates that about 2 million scholarly papers in science and engineering are published ANNUALLY in about 72,000 journals. The vast majority are read by a few hundred people at most. And this figure may not include conferences! With the increasingly interdisciplinary nature of science and engineering research, you haven't a prayer of checking every possible source. All you can do is a reasonably thorough job of checking what you think are the most likely sources. The time to stop is when you start running into diminishing returns or run out of steam. Even then, keep an eye open. I've still been finding papers related to my thesis, long after my first literature search was finished... That's why we call it "The Literature Search 'Till the End of the Universe!"