Skip to Main Content

Law 792-GRD: Legal Research and Writing for LLMs: Unit 8: Searching

UNIT 8 OVERVIEW

This page will provide you with an overview of the concepts of searching, filtering, and evaluating in legal research.

At the end of this lesson you should be able to:

  • Understand the difference between and among boolean and proximity connectors
  • Build search queries using boolean and proximity connectors
  • Articulate how various filtering mechanisms can better tailor search results
  • Demonstrate how to evaluate search results in order to select the best resource(s) for a given research question

Definitions

Boolean - a system for representing logical statements 

Natural Language (NL) - in theory, the ability of a system to understand human speech. In reality, a search mode that permits normal human language in a search query. The system processes the NL query to attempt to understand the question and return relevant results. Results of NL searches are often fixed in number, and ranked by relevancy.

Syntax - the structure and order of all the elements in your query (i.e., the order of your terms and connectors).

Boolean and Proximity Connectors

AND - A Boolean connector for a mandatory term (e.g., fall AND sidewalk will only find cases that contain both words)

OR - A Boolean connector for an optional term, often used to connect synonyms (e.g., slip OR fall OR trip will find cases that contain any of those words)

NOT - A Boolean connector for a term that should be excluded (e.g., fall NOT ice will find cases involving falls that do not contain the word ice)

AND NOT - Lexis' version of the NOT connector (note: Lexis wisely suggests that a NOT connector be at the end of your search)

& - alternate way to express the AND connector

% - alternate way to express the NOT / BUT NOT connector in Westlaw

/n - A proximity connector for a mandatory term that must appear within n words of the first term. (e.g., slip /5 ice will only find cases that contain a phrase that contains both terms separated by 5 or fewer terms). Note that /n can be replaced with /s to restrict to the same sentence, or /p to restrict to the same paragraph.

near/n - A proximity connector for a mandatory term that must appear within n words of the first term, and the order of the terms is flexible.

Additional Resources

Legal Research in a Nutshell, pp. 14-22 (General)

Legal Research in a Nutshell, pp. 74-81 (Cases)

Legal Research ina Nutshell, pp. 102-106 (Statutes)

Simple Searching

A client falls on icy steps leading into a bookstore, and injures her ankle. You want to find cases that will help you build an argument for holding the bookstore liable for her injuries (and cases that will help you counter any arguments to the contrary).

How do you search? How many of those specific terms are necessary?

  • fall
  • icy
  • steps
  • bookstore
  • injury
  • ankle
  • liable

A search that you might use just to get started, is:

is bookstore liable for injury from fall on icy steps?

This includes most of the terms from the prompt, but doesn't account for any synonyms or concepts such as "premises liability." Be prepared to add or remove terms from your search, depending on results. (hint: the first term to consider removing?  "bookstore")

How do legal research systems interpret this search query? If you don't know the answer, you're ceding all control of your search to the system.

Results in all states:

Westlaw: 53 
Lexis: 10,000+
Bloomberg Law: 1

Why such a stark difference in results?  Surely it's not because Lexis has thousands more cases on this issue than Westlaw.

Westlaw: Preferences (bottom of the page) allow you to choose to always run certain searches as Boolean Terms & Connectors:

Westlaw dialog box: Always run these searches as Boolean Terms & Connectors: Searches containing 'AND', 'OR', & or quoted phrases. (not checked). Searches containing only a single quoted phrase. (checked).

If we run the search as Boolean:  bookstore AND liable AND injury AND fall AND icy AND steps, we get the same number of results.  It appears that Westlaw processed our search as a Boolean search, putting the connector AND in between each of our terms.

Lexis: Lexis' preferences are found at the top of the page, as "Settings". But in Lexis, it's the "help" screen that reveals that Lexis Advance "automatically determines what type of search to perform from your search terms."

Lexis dialog box: "How do I search using Lexis Advance?" There are several ways to perform legal research with Lexis Advance. To begin your research, select Client at the top of the page to associate your research with a specific client or matter. Then search using one of the methods below. When you enter a search, you don't need to enter your terms using terms and connectors or natural language guidelines because Lexis Advance automatically determines which type of search to perform from your search terms. For example, if you enter budget and deficit, Lexis Advance will run a terms and connectors search, because the connector "and" is part of the search.

Lexis processed this as a natural language search, casting a very wide net, and sorting the results by relevancy.

Bottom Line:  It's important to know how each system is going to treat the syntax of your search. Check preferences/settings and help if you have any questions about how the search is being interpreted.  Or raise a question in class!

Searching for Overlapping Concepts: Building Boolean Search Queries

Nearly all searches will involve multiple facts and/or concepts. Our slip and fall example involves at least three different concepts, and we need cases that include them all:  a slip/fall injury, on icy or slippery surface, on a business' property.

With a Venn diagram, you can see how these three concepts overlap in multiple ways:Venn Diagram with three overlapping circles: slip & fall injury, icy/slippery, and business premises.

At each of the four overlapped areas, we find a different set of cases:

  • At the overlap between "slip & fall injury" and "icy/slippery", we'd find cases about falls on city streets, or on a neighbor's sidewalk.
  • At the overlap between "slip & fall injury" and "business premises", we'd find cases about falls in grocery store aisles left wet, for example.
  • At the overlap between "icy/slippery" and "business premises", there's no injury -- so no liability!
  • Finally, at the overlap between all three concepts, we'd find the cases we're looking for.

Your searches need to be constructed with these overlaps in mind.

Boolean Searches:

When you build a search query, you're telling the system which concepts are critical for you. 

Begin with a short query, and build by adding connectors. The "OR" connector is permissive: use it to separate synonyms or related concepts, of which you'd accept either.  The "AND" connector is a requirement that the term that follows be included in all results.

(slip OR fall) AND (business premises) AND (icy OR slippery)

This example allows for cases to use the term slip OR the term fall to describe the incident. It requires that cases use the phrase "business premises". And then it requires that cases use either the term icy OR the term slippery.

Once you see your results, you can (should) consider revising the search with more ANDs to reduce the number of results, or more ORs to increase the number. A better search might alter the middle set of terms to (business or store or premises), for example. You may need to repeat this process several times, especially throughout your first year. 

Building an effective search query requires an understanding of the language likely to be used in the documents you seek. If you have a limited understanding of premises liability issues, for example (perhaps not even being familiar with that term), it may take more iterations of your search query.