Finding Relevant Information
By ai-depot | June 30, 2002
Research Papers
Overview
Research papers are often released for conferences, either in the proceedings beforehand, or in the post-publications. It is often difficult to find out what conferences are relevant for the particular problem you are interested in, so I’ll give you an outline how to do this.
When you read a book or online article, there will often be some references at the end. These can be regrouped in the bibliography. The references are pointers to specific sources of interest relating to the topic. They usually have the format:
[Forbus95] Forbus, K., “Qualitative Reasoning”, CRC Handbook of Computer Science and Engineering, CRC Press, 1995.
This may not mean much to you at first, so here’s a simpler template:
[id] author, title, source, date
The id doesn’t mean anything specific, but is used within the text as indicator. The other fields are pretty much self explanatory.
Web Resource
There is a good online database that is extremely valuable. It’s called the “NECI Scientific Literature Digital Library” Research Index, and can be found at http://citeseer.nj.nec.com/cs (a.k.a. citeseer). It’s the only generic database I know of, although there are numerous smaller specific ones. You’ll find out about these as you search the web for papers, or via your local community.
It works on the same principle as a search engine, so you simply type keywords and it’ll return a list of papers. You can order your search by date to get the most recent papers. There are also cached copies of papers that you can download, which is often a better alternative than getting it from the original site. You can click on “Document Search” to limit your searches to downloadable documents.
Search Engines
The online databases may not always have what you want, or you may not know where to find the most relevant database for you. In this case, your average search engine is more than appropriate.
I’ve been using Google (http://www.google.com/) for a few years now, and it’s not let me down. That said, Wisenut (http://www.wisenut.com/) is getting more and more of my focus, since it seems to do the job terribly well.
The key in search engines is to find the most appropriate search term. If you’re looking for a specific paper, type the whole title within quotes. This works wonders, and if it doesn’t return any searches, chances are you’ll have to go to a local library.
Google: +“A Hierarchical Distributed Planning Framework for Simulated Battlefield Entities”
If you’re not looking for a specific paper, pick a set of keywords that you think is the most appropriate. You may find a few pages of interest, and read them even if they are not research material. This will reveal some more appropriate keywords that you should use for your search instead. I like to call this “iterative refinement search”, where you get closer and closer to what you were looking for by using more precise search-terms.
Viewing Papers
You’ll need a few basic tools to view papers. These are often in postscript (PS) or portable documents (PDF), and require their own viewers.
- For .PS files, you need GhostView. It works on many platforms, and the online documentation is more than adequate. You can find it right here.
- For .PDF files, you may also need Acroread. Once again, it is very portable, so you won’t have too much trouble getting it to work. Surf over here to find out more.
Reference Surfing
Once you have found one paper of interest in the field, you can easily find many others by looking at the References section. This will allow you to find all the previous papers. However, it is often useful to look for papers that are more recent. Citeseer allows you to do this, by providing a “cited by” section for each paper. This is a series of links that points to more recent papers that build upon the work you are looking at. I call this “Reverse-reference surfing”, and it can produce the most productive sessions.
Real Libraries
Citeseer’s database is quite comprehensive, but it won’t always have the papers you’re looking for. In this case, you will have to go to your local library… scary thought, I know! All you need to know is the name of the journal that the paper what released in, and ask the library attendant where you can find it. Then browse through all the volumes and pick out the one you need.
That said, few local libraries will be complete enough to have computer science and artificial intelligence information, so you may have to wander into a University library. You can usually do that for free if you don’t borrow anything, but remember to ask.
Tags: none
Category: essay |