Literature Search Strategy with AI

1 A literature search begins with clear concepts. Instead of searching with one long sentence, researchers break the topic into two to four concepts and connect them using AND. For each concept, they create a group of synonyms connected by OR. This approach improves recall while keeping the topic focused.

2 AI can help expand keywords by suggesting synonyms, related phrases, spelling variants, acronyms, and common terminology used in a field. However, keyword lists must be checked against real results. If a term yields irrelevant items, it should be removed or used only in a specific field such as title or abstract.

3 Boolean logic is essential. Parentheses control grouping, quotation marks support phrase searching, and truncation can capture variants (e.g., learn*). Databases differ in syntax. Scopus often supports field searching (e.g., TITLE-ABS-KEY), while Google Scholar is less precise but good for broad discovery.

4 Inclusion and exclusion terms should be used strategically. Early in a search, exclusions can remove relevant studies by mistake. A safer method is to run a broad search, inspect irrelevant clusters, then add NOT terms to remove those clusters. Document each change so your search is transparent and repeatable.

5 A high-quality search strategy is iterative. Researchers adjust keywords, add phrase forms, restrict to fields, and apply filters (date, document type, language). The goal is not “one perfect query,” but a set of well-documented queries that balance breadth (recall) and focus (precision).