Tuesday, May 15, 2012

Terms and Connectors Searching

Last time, I talked about the “mystery” of putting words into a search box, and the database returning documents that it thought you would want.

There is a way to override this algorithm, and in effect supply your own algorithm for finding documents. This is called "terms and connectors" searching. Describe the characteristics you are looking for in a document, and the database will return every matching document, no questions asked.


The “terms” are the words or phrases for which you are searching. These can be a mix of legal concepts and facts related to your fact pattern. Usually a database will give you a way to specify a phrase, like “dead cat bounce,” which has a much different meaning as a phrase than as three individual words.

Another way to manipulate terms is to use wildcards and root expanders. For instance, if you are looking for cases involving drunk driving, you might search for cases that mention someone being intoxicated. You would also want cases that mention intoxicating beverages. Don’t forget cases that refer to a level of intoxication. You could run different searches for each of those words, but many systems, including Westlaw and Lexis, will allow you to use a root expander and search for “intoxicat!” with the exclamation mark representing any number of letters after the phrase. Thus, it would include words like intoxicate, intoxicated, and intoxicatesque. (No, that is not a real word, but it doesn’t matter! If it were ever used in an opinion, the database would find it for you.)


George Boole, of "Boolean logic" fame
Rarely will you ever search for just one term.  Consider a fact pattern with an intoxicated bicyclist. Searching for either just “bicycl!” or “intoxicat!” is going to return many cases that aren’t really on point with what you need. How can you create a search using multiple search terms? Connectors! Databases use Boolean logic to assess your query. The most common connectors are AND, OR, and NOT. Note that while these concepts are constant across databases, the actual syntax used to make the query, like using an ampersand in place of the word AND, can vary.


If you want cases that are about intoxicated bicyclists, you probably want cases that use both terms. Connecting them with a simple AND will return only those documents with both terms appearing at least once.


“Intoxicat!” is a good way to find variants of the word “intoxicated,” but it still won’t find the word “drunk.” A case may use either word to refer to a bicyclist’s state of inebriation, so you would want to find cases with either. For this, use the OR connector, and the database will return anything with either term. This is a typical way to search for terms with common synonyms. Connecting terms with OR almost always returns more results than connecting terms with AND, and it never returns fewer results.


Occasionally, one term will be strongly associated with several other terms, like "helmet" with both "bicycle" and "football." If you find yourself getting too many cases about football helmets when you search for “helmet,” you can specify the results NOT to return cases with the word “football.” In this case, that would probably not eliminate any useful cases, but be mindful of unintended consequences when using NOT.

Proximity (within same sentence, paragraph, one word before the other, etc.)

Some databases, including Westlaw and Lexis, allow you to search for cases by the relative positioning of two terms. For instance, you can search for cases where “bicycl!” is in the same sentence as “intoxicat!” That would indicate that the document is more likely to be about an intoxicated bicyclist than one that mentions intoxication once and then a bicyclist twenty pages later.

How to do a Boolean Search

With newer systems, like WestlawNext and Lexis Advance, it can be hard to find exactly where to do a Boolean search. WestlawNext now considers it an "advanced" search. The search format is different than in traditional Westlaw too. (see below)

You can also use all of WestlawNext's "Connectors and Expanders" by starting a search in the Universal Search box (the search box at the top of every screen) with "advanced:" and then your search in parentheses.

In Lexis Advance, you can put Boolean terms directly into the search box, and the search will be treated as a terms and connectors search. Look at the search tips for the proper syntax.

The list of documents returned for any of these searches may be chronological or it may be alphabetical by something like a party name or section title. WestlawNext defaults to ranking the list by relevance using its own algorithm, but you can change that default.

Wrap up

With experience, both the natural language and "terms and connectors" methods of searching will become second nature. Law school is the best time to gain this experience, when you have such economical access to legal databases. And no matter how the syntax and algorithms behind searching changes, or varies between databases, knowing the fundamental principles behind retrieving information from legal databases will serve you well throughout your career. 

No comments:

Post a Comment