Good keyword searching practice varies from database to database, depending on how the data is structured and how the search engine works. But there are some principles that apply to most situations.
For legal research, whenever you have at least a basic understanding of the legal concepts involved, it is usually best to start out casting your net wide, and refine or expand as you go. The goal, for this strategy, is to create a search that will include ALL relevant documents and as few irrelevant ones as possible.
The following process involves an imagined scenario in which I am searching for case law on drug trafficking and entrapment. But hopefully the process is generic enough that it can apply to a variety of situations.
Step 1: Brainstorm for keywords
Think of (1) all the key facts associated with your legal problem, and (2) any legal concept that may apply. So if I am dealing with a drug trafficking case, the list of keywords might include: undercover, police, bar, drugs, suspicious, ask, request, solicit, cocaine, powder, trafficking, guilty, entrapment, dealing, selling, narcotics, controlled substances.
Step 2: Identify key concepts, group keywords
Group your list of terms into two or more key concepts. There will usually be one or more legal concepts, and one or more concepts that describe a particular factual scenario. Try to think of every possible variant (synonym) for each concept.
For example, if I want to use the entrapment defence on a drug trafficking charge, I might employ one factual concept and two legal ones:
Concept 1: drugs, controlled substances, narcotics, cocaine
Concept 2: trafficking, dealing, selling, pushing
Concept 3: entrapment
Step 3: Create search query
This is the stage where you can look at your keywords more critically, and perhaps eliminate some of them. For example, if the search engine is likely to use terminology drawn from the relevant federal statutes, then you can eliminate synonyms for “trafficking” – as this is the official name of the charge, and is certain to be used in any case on point. By eliminating synonyms, you decrease the likelihood of netting bad hits (e.g. that use a word like “dealing” in an unrelated context).
Separate synonyms using the OR operator and separate concepts (groups of synonymous terms) with the AND operator. Remember to review the search syntax (sometimes under the heading “searching tips”) for whatever database you are using. For example, is the AND operator the word “AND” or an ampersand (&) or the default (no operator). Bearing that in mind, our first search may look something like this:
(drugs OR “controlled substances” OR narcotics) AND trafficking AND entrap!
Step 4: Execute search and analyze results
Regardless of how many or how few hits you get on the first try, it is critical to analyze the results. How many hits are there? If too few, may need to drop a concept, or add some synonyms to existing concepts. If there are too many hits, you may need to add a concept, or drop some synonyms.
How are the results arranged? Relevancy? By date? Or by court level or number of cites? Can you change the order? If you cannot rank by relevancy, then take extra care to keep the number of results manageable, because the best ones might be at the bottom of the list. If you sort rank by date, consider applying a date filter to the initial search (if that is an option).
Look at a few hits – are they relevant? In the best, most relevant, results, check for key terms in the document that you may not have included in the search query. If you see a lot of poor results, check to see how and where your search terms appear in the document. Any terms that are producing poor results should be removed or modified. In some cases, consider adding a “proximity” operator, to search for two terms in the same sentence, or separated by less than a certain number of characters.
Step 5: Reformulate search
No search will net perfect results, no matter how well you refine it. On the other hand, you will rarely get optimal results on the first try. Generally, you can expect somewhere between 3 and 10 iterations to achieve a manageable set of relevant documents. But your research on this database, whatever database it may be, is still not done.
Step 6: Work Laterally
As I said, it is rare that any single keyword search will net 100% of relevant results. So now you need to read your results (the cases), and make note of references to other cases that did not come up in your search results. Then read those, and make note of further authorities. Also, for highly relevant cases, make sure to note them up, track down any citing references, and then look for citing references backward and forward in those (if they are on point), and so on.
I think of mediated searching, using a search engine or classification system, as “vertical” (think drilling down), and searching using cross-references from other relevant documents as horizontal or “lateral.” Some research tasks call for predominantly one strategy or the other, so a good researcher will know when and how to employ either strategies, or the two in combination.
Step 7: Stop
Know when to stop. You are finished your research when you either (1) stop seeing citations to unknown cases commenting on your topic (in other words you’ve seen them all, or very close to it), or (2) see repeated references to a small number of higher court authorities on which you can rely.
[This tip by Ken Fox originally appeared on the Law Society of Saskatchewan Library’s Legal Sourcery Blog]