Boolean and Natural Language Search on Trellis: A Practical Guide

Overview

Trellis supports two distinct search methods: Boolean search and natural language search. Both are available on Smart Search, and both surface results from the same underlying dataset of state trial court records. But they work differently, return different results, and are better suited to different stages of the research process.

This guide explains how each method works, when to use each one, and how to combine them to get more complete results — particularly when researching legal issues where courts and litigants use inconsistent vocabulary across filings and jurisdictions.

The core problem both methods are trying to solve: the same legal claim can be described using many different terms. A search that feels comprehensive can still miss large numbers of relevant cases simply because of the vocabulary used. Understanding how each search method handles this problem is the key to getting better results on Trellis.

The Vocabulary Problem in Legal Research

Legal concepts rarely have a single authoritative vocabulary. Courts, attorneys, and litigants across different jurisdictions use different words to describe the same conduct, the same harm, and the same legal theory. This affects search results more than most researchers realize.

Here's a concrete example: a search across New York torts cases for premises liability claims involving a physical hazard returns dramatically different result counts depending on the exact phrase used.

Every phrase in that chart describes the same basic harm. The result counts vary by a factor of more than 670. A researcher who searches for one phrase and stops there may be working with a fraction of the available relevant cases without knowing it.

Jurisdiction adds another layer of complexity. Court vocabularies don't look the same everywhere, and a search strategy that performs well in one state can miss large volumes of relevant filings in another.

In California, complaints in Los Angeles Superior Court routinely use the phrase slip and/or trip and fall as standard pleading boilerplate. A Boolean search for "slip and fall" won't surface those filings — not because the cases aren't relevant, but because the exact phrase doesn't match. What looks like a litigation volume difference between states is often a drafting convention difference.

Both Boolean and natural language search address this problem, but in different ways. Knowing which to use — and when — is what this guide covers.

How Each Search Method Works on Trellis

Boolean Search

Boolean search executes instructions precisely. You specify the terms, operators, and structure of your query, and Trellis returns results that match those exact specifications. Supported operators on Trellis include:

AND — both terms must appear in the result
OR — either term may appear
AND NOT — excludes results containing a specified term
" " — quotation marks require an exact phrase match
* — wildcard expands results to include word variations (negligen* returns negligence, negligent, negligently)
" "~n — proximity connector requires two terms to appear within n words of each other
judge:lastname, party:"name", county:countyname — field-specific operators

Boolean search is repeatable and documentable. Run the same query twice, get the same results. That makes it the right tool for applying precise criteria, narrowing a large result set, and creating a research methodology you can explain and defend.

The tradeoff: Boolean search finds exactly what you ask for and nothing you don't. If a relevant filing uses different terminology than your query, it won't appear in your results — even if it's directly on point. A query for "failure to warn" will miss filings that describe the same theory as failure to post warning signs, inadequate warning, or dangerous condition left unmarked.

🔎 For a complete reference of Boolean operators on Trellis with examples, see Using Boolean Operators in Smart Search.

Natural Language Search

Natural language search interprets the intent behind your query rather than matching exact terms. You describe a fact pattern, legal issue, or outcome in plain English, and Trellis surfaces results that reflect the legal concepts associated with that description — including results that use different vocabulary than you did.

This makes natural language search useful for two specific situations:

When you don't yet know the right vocabulary. If you're researching an unfamiliar issue or a new practice area, you may not know how courts characterize the claim, what causes of action typically appear, or what terminology is most common in the filings. Natural language search surfaces that vocabulary so you can use it.
When you want to catch related concepts automatically. Natural language search recognizes doctrinal relationships. A query for motion to dismiss will also surface results involving demurrer, because the system understands they are conceptually related. You get the complete picture without having to construct separate searches for each term.

On Trellis, natural language search and Boolean search work together — results from a natural language search can be used as a starting point for further refinement, giving researchers the breadth of concept-based search alongside the precision of keyword-based control.

🔎 For a full overview of natural language search on Trellis with example queries, see Natural Language Search in Trellis.

How to Use Each Method on Trellis

When to Use Natural Language Search

Use case 1: You know the facts but not the legal terminology.

Start by describing what happened in plain English. Trellis will surface the cases, filings, and rulings that reflect how courts and litigants characterize that situation — revealing the legal theories, causes of action, defenses, and terminology you need to build a stronger query.

Example: a shopper is detained by store security and accused of shoplifting. You know the facts. You may not yet know how courts in your jurisdiction characterize the claim, what defenses typically appear, or what the most common causes of action are. Describe the situation:

The results surface relevant filings and reveal the legal vocabulary courts use to describe this claim, including shopkeeper's privilege, false imprisonment, and merchant detention. You now have the terminology to build a precise Boolean query.

Use case 2: You need cases with a specific procedural outcome.

When you need cases with a particular result — a denied motion, a granted summary judgment, a specific ruling — you can describe the outcome you're looking for in plain English. Trellis surfaces relevant results across the full range of terminology courts use to describe that outcome, giving you a broader starting point than a single keyword query would.

This also demonstrates how natural language search handles doctrinal overlap. A query for motion to dismiss surfaces demurrer results as well, because Trellis recognizes they are doctrinally equivalent in California practice. You don't need to run separate searches for each term.

When to Use Boolean Search

Use case: You need to target a specific legal theory or validate your result set.

Once you know the vocabulary, Boolean search gives you precise control over what's included in your results. This is especially useful when you need to isolate a specific theory that applies to some — but not all — cases in a broader result set.

Example: you're researching premises liability in retail settings. A natural language search returns a broad mix of relevant cases involving wet floors, dangerous conditions, and general negligence. But your case turns on failure to warn specifically — a theory that appears in some of those filings but not all of them.

Refine with Boolean to pin down that specific theory:

You now control exactly which theories, terms, jurisdictions, and outcomes are included. The result set is scoped to your actual research objective — and you can document and reproduce the methodology.

Using Both Methods Together: A Step-by-Step Workflow

Natural language search and Boolean search work best when used in sequence. Here's how to move between them effectively on Trellis.

Step 1: Open Smart Search at trellis.law/search and select natural language search mode.
Describe the facts, legal issue, or procedural outcome you're researching in plain English. Select your state and run the search.

Step 2: Review the results for vocabulary and case patterns.
Look at the cases, filings, and rulings that surface. Note the legal theories, causes of action, and terminology that appear most frequently. This is your terminological map — the language courts and litigants actually use to describe the issue you're researching.

Step 3: Use your results to sharpen your search.
Review the cases, filings, and terminology that surfaced. Are there specific theories — like failure to warn or negligent maintenance — that appear in the results and belong in a more targeted query? Use what you've found to inform a more precise Boolean search.

Step 4: Switch to Boolean search and build your targeted query.
Apply the terms, theories, and vocabulary you've identified to a Boolean query. Add precision using the operators listed above — combine terms with AND, exclude irrelevant results with AND NOT, use wildcards to capture variations. Run the query to generate a targeted, reproducible result set.

🔎 To switch between natural language and Boolean search on Trellis, use the search mode toggle in the Smart Search bar at trellis.law/search.

Quick Reference: Which Method to Use When

Situation	Recommended Method	Why
You know the facts but not the legal terminology	Natural language search	Surfaces vocabulary courts and litigants actually use, which you can then apply in Boolean
You're researching an unfamiliar practice area	Natural language search	Reveals causes of action, defenses, and case framing before you build a precise query
You want cases with a specific procedural outcome	Natural language search first, then refine with Boolean	NLS gets you to relevant results fast; Boolean locks in the specific theory or outcome
You're researching across jurisdictions and unsure of local drafting conventions	Natural language search	Catches terminology variations and boilerplate phrases that a fixed Boolean query would miss
You know the key terms and need a precise, repeatable result set	Boolean search	Predictable results you can document, reproduce, and defend
You need to isolate a specific theory within a broader result set	Boolean search	Explicit control over exactly which terms, theories, and jurisdictions are included
You need to document your research methodology	Boolean search	Same query produces same results every time; easy to record and explain