Using AI to Develop a Search Strategy
đź’ˇ Core Idea: AI can help you figure out the language of your topic so you can figure out what other people are saying about it.
In this learning module, you will learn how to:
- Generate synonyms, disciplinary terms, and related concepts
- Translate natural language into database-ready keywords
Real talk: Generative AI is not a search engine.
I know. With a text entry box, it is deceptively similar in appearance to a search engine. While it is not designed to find sources for you, generative AI can help you design your approach to the powerful search engines that are library databases.
More real talk: Library databases are not the easiest tools to use.
With great power comes great complexity, apparently. So it makes sense to harness the conversational strength of generative AI tools to tackle database searching.
Why do we need a search strategy?
Databases don’t think like people; they think in controlled vocabulary and patterns. They usually take your searches quite literally. Unlike Google, they won’t ask, “Did you mean this instead?”
As a result, searching a database isn’t just typing words into a box—it’s also learning the words people use to talk about an idea. You say “gasoline,” I say “petroleum,” but we’re talking about the same thing. Let’s say half the world uses “gasoline”: if you search “gasoline” in a library database, it would not pull up sources using “petroleum.” You would be missing half of the information!
But how can you possibly know all the words that other people use for your topic?
Generative AI tools are designed for this very thing!
How do we design a search strategy?
It starts with your research question (link: click here to learn more about using AI to develop a question).
For example:
How does social media influence how people form their identities?
Rather than putting that question directly into an AI tool for slapdash results, let’s engage in a conversation about how to find different writers’ proposed answers to that question.
We’re going to start by asking the AI tool how to break down our topic into parts.
What key concepts are usually discussed when people write about social media and identity?
The AI tool may respond with some words or phrases:
- social media platforms
- self-presentation
- identity formation
- online communities
- adolescence
- young adulthood
- culture and belonging
Look at all these subtopics that fall under our main topic! It’s clear that research questions are made of many parts, not single ideas.
Now we can ask for synonyms and related terms to figure out what words people from different perspectives use to talk about similar ideas.
What are alternative terms or phrases people use for identity formation?
This prompt can obviously be repeated with any of the subtopics previously identified.
The AI tool might respond with a list like this:
- self‑concept
- self‑identity
- identity development
- sense of self
- personal identity
Then we can ask the AI tool to form a search strategy by combining terms using Boolean operators.
Use terms from across these concepts to create database search strings using Boolean operators.
You can specify which concepts you are more interested in to better focus the search strings.
🔎 Search strings are carefully formed prompts designed to communicate clearly with the controlled vocabulary and algorithms of a library database.
The AI tool may give you something like this:
("social media" OR Instagram OR TikTok OR "online platforms") AND (identity OR "self-presentation" OR "sense of self")
This is an example of quite a complex search string.
🎓 PRO TIP: Instead of searching for sources that perfectly address all parts of your research question, think of your search strategy as putting together the pieces of a puzzle. Search for one or two concepts at a time, and repeat with different combinations. Each search may yield a single puzzle piece. Your job as the writer is to bring the pieces together into one cohesive whole.
With this example, you might search for “social media platforms” and identity, find a couple promising results, then conduct a separate search for “social media platforms” and adolescence, and find a couple more results.
But wait—there’s more!
Treat your first search as a rough draft. When you type or paste the search string into the database, take a moment to evaluate the search results. Are they too broad? Too specific? Focused on the wrong thing?
Report back to the AI tool with your concerns and ask for ideas to refine the search string.
I’m getting too many irrelevant results. Refine the search string to focus on college students.
Try out lots of search strings! If you think you’ve tried enough, try one more.
🎓 PRO TIP: As you look at search results on a library database, notice the subject headings that are listed with each result. These headings can provide insight into what terms are common for the topic.
For example, you may discover that scholars use “digital identity” rather than “online self.” Report this finding to your AI tool so it can help you revise the search string.