Notes on Library Databases, Keyword Search Strategy, and AI Considerations in Research

Context and session objectives

• This session focuses on research database practices, comparing how search works in Google/AI chat tools vs library databases.
• Homework reminder: watch the video from the previous week; purpose is to learn proper database search techniques.
• Key takeaway: library databases require keyword-based searches, not full-sentence prompts, and rely on structured filters to refine results.

Key databases introduced

• JSTOR (recommended starting point for this class):
  • Located under the JSTOR portal (often abbreviated as j.).
  • Strengths: rich in subject-specific articles; full-text PDFs are typically available.
  • Abstracts appear at the beginning of entries; sometimes include outer keywords or subject terms.
• Academic Search Complete (ASC):
  • Listed as another core database to know (top of the database list).
  • More suitable for contemporary topics; tends to return a broader, often more current set of sources.
• Overall strategy: don’t memorize every database; focus on knowing JSTOR and Academic Search Complete and learn the workflow common to most databases.

Keyword-based search vs full-sentence prompts

• In databases, you construct queries using keywords rather than full sentences.
• Example topic: voting rights and the United States.
• Possible initial query structure (keywords-based):
  • Simple: women AND voting rights AND United States
  • Synonyms: include voting rights OR suffrage to cover terms used across sources
  • Grouping with parentheses to signal equivalence of terms: use OR inside parentheses
• Boolean logic concepts demonstrated (in practice you’ll apply them in the interface):
  • AND (main connectors between concepts)
  • OR (to include synonyms or related terms)
  • Parentheses group terms to control evaluation order
• Example boolean expression (formatted in LaTeX for clarity):
  $( ext{women} \land ext{United States} \land ( ext{voting rights} \lor ext{suffrage} ) )$
• Important nuance from the lecture: the order of operators matters in most interfaces; AND is used to narrow, OR is used to broaden within grouped terms.

Building and refining a search query (step-by-step workflow)

1) Start with simple keywords related to the topic:

• Example: women, voting rights, United States
  2) Add synonyms for key concepts to capture variations:
• Example: (voting rights OR suffrage)
  3) Use grouping to combine concepts coherently:
• Example: (women AND United States AND (voting rights OR suffrage))
  4) Consider abstract-level searching to improve relevance:
• Use Modify/Search options to restrict your search to the abstract field so that the concepts appear in the article abstract.
  5) Check results and refine keywords based on what appears:
• If results are too broad or irrelevant, revise terminology, synonyms, or narrow to specific fields.
  6) Apply filters to narrow results further (left-hand panel in most databases):
• Document type: filter to journal articles, book chapters, etc. (choose what’s most relevant)
• Date range: tailor to the period of interest (e.g., post-2000 for modern topics)
• Language: English, Spanish, etc., if needed
• Source type: peer-reviewed or other scholarly sources
• Full text availability: ensure you can access the full article
  • After filters are applied, click Apply (or equivalent) to finalize the narrowed results.

Abstracts, filters, and sample results

• Abstracts: the summary at the beginning of an entry; helpful for quickly judging relevance.
• Practical impact of using abstract filter: dramatically reduces results from a large pool (e.g., from about $154000$ results down to around $5540$ after adding abstract focus).
• Date filters: adjust to prefer more recent literature (e.g., since 2000) for topics like social media or AI.
• Language filters: limit to languages you can read (e.g., English and Spanish).
• Other common filters: full text, peer-reviewed, subject areas, geographic location.
• JSTOR note: across JSTOR you typically have access to full text PDFs, which can simplify citation and reading.

Using filters effectively across databases

• Most databases share a similar workflow: keywords-based search, then filters on the left, then Apply.
• If results are too numerous, refine by:
  • Narrowing by abstract or keyword fields
  • Adding date and language constraints
  • Selecting only journal articles if needed (as opposed to chapters or other formats)
• When results are still too broad, revisit your keywords and synonyms rather than piling on filters.
• The professor emphasized starting with JSTOR for robust subject coverage and full text access, then using ASC for more contemporary topics.

Practical takeaways and recommendations

• Always start with a simple keyword set and build up with synonyms using OR within parentheses.
• Use AND to combine major concepts (e.g., topic, location, timeframe).
• Validate results by inspecting abstracts and applying targeted filters to improve relevance.
• JSTOR is a strong starting point due to its full-text PDFs and subject depth; ASC is valuable for more current materials.
• If you struggle, use the chat or instructor support to troubleshoot search strategies.
• Remember: the quality of results hinges on keyword choice and appropriate filters; don’t rely on single broad terms.
• The in-class activity timing noted: 15 minutes for exploration, 5 minutes for setup and questions; manage time accordingly.

AI tools in research: opportunities and cautions

• AI chat tools (e.g., ChatGPT, Gemini, Copilot) can provide quick drafting and idea generation but have notable caveats:
  • They can produce plausible-sounding but not always accurate information.
  • They may reproduce or invent citations that do not exist (fake citations risk).
  • Sign-up and use may involve data privacy concerns, as many services may capture and analyze your searches and profile data (you are effectively the product).
• Practical guidance from the session:
  • Treat AI as a starting point for ideas and a way to check framing, not a substitute for original search and critical evaluation.
  • Always verify AI-provided information against primary sources and bibliographies.
  • Use AI to draft or outline, but supplement with direct source checking and literature review.
  • If you rely on AI for citations, be prepared to verify citations; several fake citations can arise, especially with newer literature.
• Ethical and professional implications:
  • Outsourcing writing to AI may impede learning and skill development.
  • In a competitive job market, foundational research skills remain essential.
  • Some claims from AI may be biased or misrepresent sources; cross-check with authoritative references.
• Example demonstration discussed:
  • A long paragraph produced by AI about women in WWII workforce and child impacts; the AI-generated text contained some inaccuracies or unsubstantiated claims (e.g., a claim about emotional neglect not necessarily supported by all sources).
  • This illustrates the risk of taking AI output at face value without source verification.
• Best practice suggestion:
  • Use AI to brainstorm, generate outlines, or identify potential search terms, then search for primary sources to verify and cement your argument.
  • Use library databases to locate authoritative sources and citations rather than relying on AI-generated content alone.

Verifying sources and avoiding fake citations

• The session included a quick exercise: search for citations 1–4 in a library/catalog or Google Scholar to see if they exist.
  • The example showed that some citations may not exist, illustrating the risk of relying on AI-generated or misattributed references.
• Real-world concern: fake citations have circulated in recent years (notably around 2024–2025) due to AI-assisted text generation and citation fabrication.
• Verification strategies:
  • Cross-check citations against library catalogs (e.g., WorldCat) and the publisher’s site.
  • Confirm that the source actually exists and is accessible via your university resources.
  • Favor primary sources and peer-reviewed sources when possible.
• Caution about paywalls and training data:
  • Much scholarly content is behind paywalls, while AI training data may pull from open web content.
  • The discussion notes that relying on free AI tools for writing or citation can undermine access to rigorously vetted sources.
• Practical takeaway:
  • Do not rely on AI-generated citations; always verify with library resources and primary sources.
  • If an AI tool provides a citation, locate the source yourself and confirm its details before including it in your work.

Real-world implications and classroom guidance

• Emphasize that a strong literature review relies on careful search design, critical evaluation, and proper citation practices.
• Use JSTOR for depth and full-text access; use ASC for breadth and current topics.
• Be mindful of ethical considerations when using AI tools in academic work: do not outsource your core writing; use AI as a support tool and verify all outputs.
• If you encounter uncertain citations or questionable sources, consult library staff or the course instructor for guidance.
• When in doubt about a source’s credibility, check for:
  • Peer-review status
  • Publisher credibility
  • Relevance to the topic
  • Recency and cross-citation patterns

Quick reference: practical examples you can apply now

• Example 1 (JSTOR query):
  • Query: $( ext{women} \land ext{United States} \land ( ext{voting rights} \lor ext{suffrage} ))$
  • Start simple, then adjust keywords based on results.
  • If results are too broad, add filters or focus on abstracts to improve relevance.
• Example 2 (Abstract-focused narrowing):
  • Adjust the search to include a term in the abstract to ensure it appears in the article’s summary, drastically reducing results.
  • Result evolution (illustrative): from $154000$ to roughly $5540$ results after abstract filtering.
• Example 3 (ASC for contemporary topics):
  • If your topic is very current (e.g., AI, social media), ASC may yield more up-to-date sources.
• Example 4 (Citation verification exercise):
  • Look up a set of four citations in your library system or WorldCat to confirm they exist before including them in your bibliography.

Summary takeaways for exam preparation

• Master the keyword-based search workflow: start with simple terms, incorporate synonyms with OR, group with parentheses, and combine concepts with AND.
• Use abstracts and targeted filters (date, language, full text, peer-reviewed) to refine results efficiently.
• Begin with JSTOR for depth and PDFs; turn to ASC for breadth and up-to-date coverage.
• Treat AI as a supplementary tool: it can aid brainstorming and drafting but is not a substitute for rigorous source verification and scholarly writing.
• Be vigilant about citation accuracy: verify every citation with authoritative sources and library catalogs to avoid fake or unverifiable references.
• Always consider ethical implications and privacy when using AI tools; ensure your learning remains the primary goal of scholarly work.