Indexing, keywords, ranking, personalization, and the structural limits every researcher must understand before trusting a results page
Every research task begins with a search. Students type a question or a few keywords into a box — a library catalog, an academic database, or a general search engine — and within a fraction of a second, a ranked list of results appears. This speed creates an impression that search is effortless, exhaustive, and authoritative. It is none of those things.
This chapter opens Part IV by explaining what actually happens between a search query and a results page. Students will learn how systems build indexes, how keywords and metadata are matched against those indexes, how results are ranked, how popularity and personalization shape what different searchers see, and where search systems systematically fail to reach available information. Understanding these mechanics is not a technical detour from research — it is the foundation for every search skill that follows in this part of the textbook.
Students who treat search as a black box accept whatever appears first. Students who understand how search works take deliberate control of their results, search multiple systems with purpose, and evaluate every source on its own merits rather than trusting its position in a list.
Two students research the same question: "Does social media use affect teenage mental health?" The first student searches Google, opens the first three results, and begins writing. The second student searches the same question in a web search engine and an academic database, compares the kinds of sources each system returns, evaluates several results before choosing any, and records the vocabulary that produced the most useful results. Both students spend the same amount of time searching. Which student is more likely to find strong evidence — and why? By the end of this chapter, you will be able to answer that question precisely, and you will understand the reasoning behind the answer well enough to apply it to any search task you encounter.
When students type a query into a search box and press enter, what happens looks instantaneous. Results appear in fractions of a second. This speed creates an impression that search is effortless, exhaustive, and authoritative. It is none of those things.
Search is a matching process. A system compares the words in a query against an index of stored information and returns results ranked by some definition of relevance. The speed is real. The authority is not.
| Common Assumption | What Is Actually True |
|---|---|
| Search reads everything in real time | Search consults a pre-built index. Unindexed documents cannot appear, no matter how relevant or high quality they are. |
| Search understands meaning like a person does | Search matches patterns of characters and learned statistical associations. It can mistake one meaning of a word for another and miss context a human reader would catch. |
| Search results are complete | No system indexes everything. Databases index only what vendors license; search engines reach only part of the public web. |
| High ranking means high quality | Ranking reflects many signals, most unrelated to accuracy, credibility, or scholarly value. |
None of this means search is useless — quite the opposite. Used strategically, search is one of the most powerful research tools available. It offers rapid, automated access to vast collections of indexed information, connecting researchers to sources across institutions, disciplines, and formats that would otherwise require extensive travel or specialized access. Used passively, however, search is one of the easiest ways to find the wrong answer efficiently.
Students who do not understand how search works often draw three conclusions that weaken their research: that the first result is automatically the best result; that if something does not appear in a search, it does not exist; and that searching harder simply means typing more words into the same box. This chapter exists to replace each of these assumptions with an accurate understanding of how search actually functions.
The goal of this chapter is to shift students from passive searchers — who accept whatever a system hands them — to strategic searchers, who understand the mechanics well enough to take deliberate control of their results. That shift begins with understanding what is actually happening when a search query is processed.
Before any search can happen, information must be indexed. Indexing is the process of reading, analyzing, and cataloging information so it can be retrieved later. Think of an index at the back of a textbook. The index does not contain the full text of every page — it records which words and topics appear where so readers can jump directly to relevant sections.
Search systems build indexes in the same spirit but at enormous scale. A web search engine maintains an index of billions of documents. An academic database maintains an index of hundreds of thousands or millions of scholarly records.
Indexing begins with a crawler or harvester. For web search engines, automated programs called crawlers follow links from page to page across the internet, collecting text and metadata. For library databases, records are ingested from publisher feeds, data-sharing agreements, or manual cataloging by librarians and indexing specialists.
Once documents are collected, the system processes them. Processing typically involves identifying the significant words in each document; stripping common words such as "the," "and," and "is" that appear everywhere and carry little distinguishing meaning; stemming words to their root forms so that "running," "runs," and "runner" are treated as versions of the same root term; and recording how often each significant term appears and where in the document it appears.
The result is an inverted index: a data structure that maps each term to a list of documents containing that term. When a query arrives, the system looks up the query terms in this index and retrieves the matching documents. This is why retrieval feels instantaneous — the system is searching a pre-built structure, not reading documents live.
Beyond the text of documents, indexing also captures metadata — structured information about a document rather than the document's content itself. Metadata in a library catalog record includes the title, author, publication date, subject headings, call number, and format. Metadata on a web page may include the page title, description, author tag, publication date, and category tags.
Metadata is often more reliable for retrieval than full text alone because it is structured and controlled by professionals rather than generated incidentally by an author's word choices. A catalog record assigned the subject heading "Climate change — Economic aspects" will reliably surface relevant documents even if those exact words never appear prominently in the document text. This is why controlled vocabulary and subject headings, covered in depth in Chapter 22, are such valuable research tools.
The quality of any retrieval depends entirely on the quality of the underlying index. A poorly built index with incomplete records, missing metadata, or outdated information will produce poor results even for a perfectly constructed query.
This is one reason library databases, built and maintained by professional catalogers, often produce more reliable retrieval for academic research than systems with looser or fully automated indexing practices.
The words a searcher chooses matter enormously. The same underlying research question can produce very different results depending on how it is phrased, because a search system breaks a query into individual terms and looks each one up in its index. Documents containing more of the query terms — particularly in prominent locations such as the title or abstract — are typically ranked higher than documents containing fewer terms or containing them only in passing.
This creates a practical reality every researcher must reckon with: if the words in a query do not match the words used in the documents a researcher actually needs, those documents may not appear at all, regardless of how directly relevant they are.
Many search systems allow users to target specific metadata fields rather than searching full text. In a library catalog, a student can search by author, title, subject, or call number. In an academic database, a student can limit a search to article titles, abstracts, or author-assigned keywords.
Searching metadata fields produces more precise results than full-text searching because it targets structured, controlled information rather than the unpredictable language of document content. An author search for a known scholar retrieves all indexed works by that person regardless of how any individual document happens to be worded. A subject heading search retrieves documents systematically assigned to that category by professional indexers, independent of whether the exact search terms appear anywhere in the text.
| Search Type | What It Targets | Best Used When |
|---|---|---|
| Full-text keyword search | Every word in the document, regardless of location | Exploring a topic broadly or when the exact vocabulary of the field is still unfamiliar |
| Title or abstract search | Words appearing in the title or summary fields only | Narrowing results once key vocabulary is known |
| Subject heading search | Controlled vocabulary assigned by professional indexers | Retrieving everything on a topic regardless of the specific wording used by individual authors |
| Author or institution search | Structured metadata identifying who produced the source | Following a known scholar's body of work or an organization's publications |
Strategic searchers use both keywords and metadata together. They begin with broad keyword searches to explore a topic, then shift to subject headings and field-specific searches to narrow and refine. They pay close attention to the vocabulary used in the most productive early results and incorporate that vocabulary into follow-up searches. They use metadata filters to limit results by date, format, language, or publication type. Understanding this relationship — between the words a searcher chooses and the words a system has indexed — is one of the most immediately practical research skills a student can develop.
Chapter 9 taught students how to build keyword lists with synonyms, broader terms, narrower terms, and related vocabulary before searching. This section explains exactly why that preparation matters: a search system can only retrieve what its index contains under the terms a searcher actually uses. A well-built keyword list is not a formality — it is the mechanism by which a research question becomes retrievable.
Finding documents that match query terms is only the first step. Search systems then rank those documents in order, presenting some results before others. Understanding how ranking works explains why the first result is not always the best result for a research task.
Ranking attempts to order results by estimated relevance — placing documents most likely to satisfy the query near the top. The definition of "relevance" differs significantly between systems, and that difference matters enormously for researchers.
Academic databases typically use relatively transparent relevance criteria. A result may be ranked higher if query terms appear in the title, keywords, or abstract rather than only deep in the body text; if all query terms are present rather than just some of them; or if the terms appear together in close proximity rather than scattered across a long document. Some databases additionally weight results by publication recency, citation count, or the standing of the publishing journal.
Academic database ranking tends to prioritize content relevance over other factors, which makes database search more predictable — and generally more appropriate for academic research — than commercial web search.
Web search engines use far more complex and far less transparent ranking algorithms. The foundational insight behind early web search was that a page linked to by many other pages is probably more important than a page linked to by few. This link analysis, combined with assessment of link quality and the text surrounding links, remains a significant ranking factor today.
Modern web search ranking incorporates hundreds of additional signals, including how often users click a result, how long users stay on a page after clicking, how quickly a page loads, whether a page displays well on mobile devices, the general authority of the hosting domain, geographic signals, recency, inferred query intent, and broader patterns of user engagement.
| Ranking Factor | Academic Database | Commercial Web Search Engine |
|---|---|---|
| Term placement (title/abstract) | Strong factor | Moderate factor |
| Link popularity | Rarely a factor | Major factor |
| Click-through behavior | Rarely a factor | Major factor |
| Citation count | Sometimes a factor | Not typically a factor |
| Advertising relationship | Not a factor | Major factor (sponsored placement) |
| Personalization | Rarely a factor | Significant factor |
A highly ranked result is not necessarily accurate, authoritative, or appropriate for a research question. It is a result that the ranking algorithm estimated would satisfy a generalized version of the query, based on many signals — most of which have nothing to do with scholarly quality. Researchers who accept rankings uncritically are allowing an algorithm to make their source-selection decisions for them.
Two additional factors shape the results students see in everyday search: popularity and personalization.
In web search, popularity is a significant ranking signal. Pages that many people have linked to, clicked on, and shared tend to rank higher than pages with less engagement. This is not inherently bad — popular pages are often popular precisely because they contain useful, accessible information. Encyclopedic reference sites, major news outlets, and well-established institutional websites often rank well because they genuinely serve large numbers of users.
However, popularity has real limitations for research purposes. Popular content tends to reflect mainstream views, dominant perspectives, and widely accessed information. Specialized scholarship, minority perspectives, emerging research, underrepresented voices, and technical expert sources may rank poorly simply because they attract less general web traffic — not because they are less accurate or less relevant. A peer-reviewed virology database is far less "popular" by traffic metrics than a wellness blog, yet it is categorically more appropriate for research on infectious disease.
Popularity also means that viral but false information can rank well, at least temporarily. During fast-moving news events, health crises, or political controversies, inaccurate content sometimes accumulates clicks and links rapidly, briefly elevating it in rankings before fact-checking organizations and authoritative sources can respond. Chapter 28 examines this dynamic in detail.
Many search systems personalize results based on a user's history. Web search engines track previous searches, clicked links, browsing history, location, device, and account activity. Over time, the system builds a model of a user's interests and adjusts results accordingly.
Personalization can be convenient in everyday life. For research purposes, it creates real problems. Research requires exposure to diverse, sometimes unfamiliar information. A student whose results are shaped by personal interest history, geographic location, or behavioral signals from past searches may see a systematically narrowed version of available information. Two students searching the identical query on the same day, from different accounts and locations, may see substantially different results.
Personalization also contributes to what researchers call filter bubbles — environments in which users consistently see information aligned with their existing patterns rather than the full range of available perspectives. Chapter 31 addresses filter bubbles and recommendation algorithms in depth.
Students can partially reduce personalization effects by searching in private or incognito browser modes; by using a different browser or device for research than for casual browsing; by searching while logged out of personal accounts; or by relying on academic databases that do not track personal behavior the way commercial search engines do.
These steps do not eliminate personalization entirely, but they reduce its influence. More importantly, students should remember that any web search result reflects a filtered version of available information — filtered through popularity, personalization, and ranking algorithms simultaneously. Academic databases, library catalogs, and specialized indexes are generally less subject to these distortions and are usually more appropriate starting points for scholarly research.
Every search system has structural limits — categories of information it cannot reach, formats it handles poorly, and types of sources it systematically underrepresents. Understanding these limits prevents researchers from making one of the most common research mistakes: concluding that absence from search results means absence from existence.
Much of the information that exists is not accessible through general web search. This category is sometimes called the deep web or invisible web. It includes several distinct kinds of material.
| Category | Why It Is Invisible to Standard Search | Example |
|---|---|---|
| Paywalled content | Requires subscription or institutional access that crawlers cannot reach | Most peer-reviewed journal articles |
| Database-only content | Presented through a private search interface, not exposed to general web crawling | PubMed, LexisNexis, PsycINFO, most library catalogs |
| Resistant formats | Handwritten, audio/video without transcripts, images without captions, underrepresented languages | Archival letters, oral history recordings |
| Never digitized | Exists only in physical form and has not been scanned, transcribed, or indexed | A 1920 newspaper on microfilm; a handwritten 1845 letter |
Academic databases are more comprehensive than web search for scholarly literature, but they have their own structural limits. Most databases are subject-specific, covering particular disciplines. A search in an education database will not return psychology literature even if the research question genuinely spans both fields. A search in a history database will not return scientific journal articles.
Databases also reflect decisions made by publishers, vendors, and librarians about which sources to include. Regional scholarship, non-English-language journals, and emerging open-access repositories may be underrepresented or absent entirely. Gray literature — government reports, white papers, think-tank publications, conference proceedings, dissertations, and technical reports — is inconsistently covered across different databases.
Indexes reflect the state of information at the time they were built. Newly published articles may take time to appear in databases after indexing. Newly created websites may not be indexed yet. A search conducted today reflects the index as it currently exists, which may lag behind the actual state of published information by days, weeks, or longer.
Conversely, old information persists in indexes long after it has been superseded, retracted, corrected, or discredited. Researchers must pay attention to publication dates and check whether information has since been updated or withdrawn.
"I searched and found nothing on this, so it must not exist" is one of the most common and most damaging conclusions a researcher can draw. Given everything covered in this section — paywalls, database fragmentation, undigitized archives, and indexing lag — absence from a single search system is weak evidence of absence from reality. The correct response to an empty or thin result set is to try a different search system, different vocabulary, or a subject librarian — not to conclude the topic has no available evidence.
Knowing these limits is not a reason to avoid searching — it is a reason to search more strategically. Students should use multiple search systems for the same topic rather than relying on just one. They should consult subject-specific databases alongside general databases. They should ask a librarian when they need access to specialized or hard-to-find sources. And they should never assume that the absence of results in one system means the absence of relevant information altogether.
A list of search results looks orderly. Items appear in sequence, labeled with titles and brief descriptions, ranked from most to least relevant. This visual presentation creates an impression of objectivity and thoroughness.
Search results are not neutral.
Every search result list is the product of a system designed by human beings with particular goals, priorities, and constraints. That system made decisions about what to index and what to leave out, how to interpret query terms, how to weight relevance signals, and how to present results. Those decisions reflect values, interests, and limitations whether or not they are explicitly stated to the user.
A commercial web search engine is designed primarily to serve users in ways that support advertising revenue. Highly ranked results tend to be results that users click on and that advertisers are willing to pay to appear near. Authoritative, specialized, or non-commercial content competes for visibility against well-funded content marketing.
An academic database is designed to serve researchers within specific disciplinary traditions. The sources it includes reflect the publishing norms, languages, and institutional prestige structures of those traditions. Journals from dominant academic institutions, wealthy countries, and established disciplines are typically better represented than emerging scholarship, multilingual journals, and alternative academic communities.
A results list represents what one particular system, built with particular priorities and limited by particular coverage, was able to retrieve and chose to rank highly for one particular phrasing of a question. Change the system, the phrasing, the date, or the searcher's account history, and the list changes. This is not a flaw to be embarrassed about — it is simply what a results list is. Treating it as a complete, neutral snapshot of available knowledge is the error to avoid.
Web search results regularly include advertising. Sponsored results appear at the top and sometimes the sides of results pages, visually formatted to closely resemble organic results. Students who do not distinguish between sponsored and organic results may click an advertisement believing it to be the most relevant research result. In a research context, sponsored content introduces a clear commercial bias that has nothing to do with evidentiary quality.
Library databases do not typically include advertising, which is one of several reasons they remain preferred starting points for academic research.
Search results are curated by algorithms, and algorithms encode the assumptions of their designers. An algorithm that weights recent content heavily will surface different results than one that weights citation counts heavily. An algorithm trained on click behavior will reflect what past users found engaging — not necessarily what is most accurate or appropriate for a specific research question.
This curation is largely invisible to the user. Students see a list and may have no idea which signals shaped it. That invisibility can create false confidence in results that have not actually been evaluated by anyone for accuracy.
Because search results are not neutral, researchers cannot outsource source selection to the algorithm. Every result requires evaluation. The MGU Source Evaluation Model — relevance, authority, evidence, accuracy, purpose, currency, and context — applies to every source regardless of where it appears in results or how highly it is ranked. Strategic researchers also notice what is missing from results, not only what is present. If a search on a controversial topic returns results representing only one perspective, that pattern itself is informative: it may mean the search terms were too narrow, the system is filtering results, or the indexed sources predominantly represent one viewpoint. Responsible research means questioning the shape of results, not only accepting them.
Understanding how search works leads directly to better search practice. Strategic searching is not about typing more or searching longer. It is about searching with intention, adjusting based on what results reveal, and knowing when and where to search differently.
| Passive Searcher | Strategic Searcher |
|---|---|
| Types the first phrase that comes to mind | Identifies key concepts and generates a keyword list before searching |
| Opens whichever search box is most familiar | Chooses a search system appropriate to the task — catalog, database, or web engine |
| Trusts the first few results automatically | Reads results critically and applies the MGU Source Evaluation Model |
| Treats the first search as the final search | Iterates — refining vocabulary, trying new systems, following citation chains |
| Gives up when results look thin | Recognizes when to escalate to a subject librarian or specialized database |
Strategic searchers think before they type. Before entering a query, they identify the key concepts in their research question, generate synonym lists, consider disciplinary vocabulary, and choose an appropriate search system for the task. They know that the same question might require a different search approach in a library catalog, an academic database, and a web search engine.
Strategic searchers read results critically. They examine the sources behind results, not just the titles and snippets. They notice who published the information, when it was published, and whether it is appropriate for the research question at hand. They do not accept high ranking as evidence of quality.
Strategic searchers iterate. An initial search is almost never the final search. Strategic searchers examine what their first results contain, identify useful vocabulary in productive results, and use that vocabulary to refine subsequent searches. They follow citation chains from useful sources to find related material, a technique explored fully in Chapter 22. They search multiple databases and systems for the same topic rather than stopping at the first system they tried.
Strategic searchers know when to stop searching alone and when to escalate. They recognize when a topic requires specialized expertise — and when that means consulting a subject librarian rather than searching harder on their own.
Most students begin as passive searchers. They type the first words that come to mind into a familiar search box and work with whatever appears. Passive searching is fast but shallow, and it consistently misses better information that strategic searching would find.
The shift from passive to strategic is gradual. It develops through deliberate practice — trying different search terms, comparing results across systems, and reflecting on why some searches work better than others. The remaining chapters of Part IV build this practice across specific systems: library catalogs, academic databases, open web research, advanced Boolean techniques, and the MGU Library itself.
Chapter 3 introduced research as inquiry beginning with questions, not answers. Chapter 9 taught students how to translate research questions into search terms. This chapter has explained what happens when those terms enter a search system. The chapters that follow in Part IV build practical search skills for specific systems: library catalogs and digital collections (Chapter 18), academic databases (Chapter 19), open web research (Chapter 20), Boolean operators and advanced search (Chapter 21), subject headings and citation chaining (Chapter 22), the MGU Library specifically (Chapter 23), and managing search results responsibly (Chapter 24). Throughout all of them, the underlying principle remains constant: search is not a passive act of receiving information. It is an active, strategic skill that researchers develop and refine across an entire career.
Think about how you typically search for information — for school, for personal interest, or for everyday decisions. Do you usually click the first result? Do you try more than one search system? Share your habits with a partner or small group and discuss: based on this chapter, which of your habits are strategic, and which might be worth changing?
Search the same topic in a commercial web search engine and in a library database. Compare the first five results from each. How do the types of sources differ? How do the ranking criteria seem to differ? Which system produced results more appropriate for an academic research paper, and why?
Choose a topic related to your intended major or an interest area. What kinds of sources on this topic might exist in the invisible web — behind paywalls, in specialized databases, or in non-digitized archives? How might you access this material if a general web search does not surface it?
If two people can search the identical question and receive substantially different results because of personalization, what are the implications for public debate, civic decision-making, and shared understanding of facts? How might this connect to disagreements you have witnessed in everyday life?
Choose a research topic you are currently working on, or select one from the following list:
Using this topic, conduct the following searches and record your results: a general web search engine search using a brief, natural-language query; a search in a library database your institution provides access to using the same topic; and a search of the MGU Library's encyclopedia or reference collection for background on the topic.
For each search, record the search system used, the exact query typed, the first five results (title and source only), how many results the system reported, and what types of sources appeared (news articles, academic journals, websites, books, reference entries, other).
After completing all three searches, write a paragraph comparing the results. Address these questions: How different were the results across systems? What types of sources dominated each system? Which system produced results most appropriate for academic research? What does this comparison suggest about choosing search systems strategically?
Choose one result from a web search engine and investigate what you can learn about how and why it appeared. Record the query you used; the title and URL of the result; where it appeared in the results list (first, third, tenth, and so on); who published it (individual, organization, institution, commercial entity); when it was published or last updated; whether it appears to be sponsored or organic; and what links to it, if you can determine this.
Write a short reflection: What factors do you think contributed to this result appearing where it did? What does this tell you about how the search system ranked it? Is this source appropriate for research? Why or why not?
Relevance: Does this source address your actual research question, or only a related topic?
Authority: Who produced this source? What are their credentials? Is this an individual, an organization, an institution, or a commercial entity? Do they have expertise in this subject?
Evidence: What evidence does the source offer? Is it based on research data, expert analysis, personal experience, or opinion?
Accuracy: Can the claims in this source be verified against other reliable sources? Are sources cited? Are there factual errors?
Purpose: What is the purpose of this source? Is it to inform, persuade, sell, or entertain? How does purpose affect trustworthiness?
Currency: When was this source published or updated? Is recent information required for this topic?
Context: What was the original context of this source? Has it been taken out of context, misrepresented, or misattributed?
The Research Notebook is where students record what they discover during a search session — not just sources, but process. After any significant search session, record the following using the MGU Research Note Model as your guide.
Example entry:
One of the most persistent misinformation risks in search-based research is the assumption that high ranking equals high trustworthiness. Students who accept the first few results of any search as authoritative — without evaluating those results independently — are vulnerable to this trap.
High-ranked results on commercial search engines may include content marketing produced by organizations with financial interests in shaping what readers believe; opinion articles formatted to resemble news reports or research summaries; aggregator pages that summarize, and sometimes distort, other sources rather than providing original information; websites optimized for search visibility through technical tricks rather than content quality; and pages that have accumulated many links because they are popular, not because they are accurate.
Apply the MGU Misinformation Check to any search result that will be used in research:
Research has shown that users click the first result in a search list far more often than any other result, regardless of whether the first result is actually the most useful or accurate. This click pattern is driven by interface design and cognitive habit, not by evidence of quality. In research, the first result deserves evaluation — not automatic selection.
When a search produces a useful source, students must immediately record enough citation information to find the source again and cite it correctly later.
| Source Type | Minimum Information to Record |
|---|---|
| Web source or online article | Author(s); title of article or page; website or publication name; full URL; date published or last updated; date accessed |
| Database result | Author(s); article title; journal or periodical name; volume, issue, and page numbers if available; publication date; DOI if available; database name |
| Library catalog record | Author(s); book title; publisher and place of publication; year of publication; call number and library location |
Students often save only a title or URL when they find a useful source, intending to retrieve full citation details later. By the time they are ready to cite, the page may have moved, the browser tab may be closed, or memory of the full details may have faded. Record complete citation information at the moment of discovery — not later.
For any source from Activity 17A that you found useful, record complete citation information now, before continuing to the next chapter.
When students understand how search works at the level of indexing, ranking, and retrieval, they use every MGU Library resource more effectively. The MGU Library is not simply a collection of links. It is a structured system for connecting researchers to appropriate, high-quality information — and using it well requires understanding how information is organized and retrieved.
Search skills are not confined to a research methods course. Choose one of your other current courses and identify a recent or upcoming assignment that requires finding information.
Answer the following in writing: (1) Which search system is most appropriate for this assignment — library catalog, subject database, or web search engine — and why? (2) What vocabulary might researchers in this field use that differs from everyday language? (3) What kind of source would you expect to rank highly in a general web search on this topic, and is that source type actually appropriate for this assignment? (4) What part of the invisible web might contain important sources for this assignment that a general web search would miss?
This activity asks you to begin thinking about search mechanics as transferable knowledge rather than as a skill specific to one class — a habit that will serve you in every course you take at MGU.
This chapter argues that researchers must "not merely find information — earn trust in information." How does understanding indexing, ranking, popularity, personalization, and search limits contribute to earning trust in what a search returns? Identify one specific habit from this chapter that you will change in how you search going forward, and explain why.
Chapter 17 explained the mechanics behind every search system: indexing, keyword matching, ranking, popularity, personalization, and structural limits. Chapter 18 applies these concepts to the first specific search system in Part IV — the library catalog. You will learn how catalog records are structured, how to search by title, author, and subject, how call numbers and classification systems organize physical and digital collections, and how to move efficiently from a catalog record to an actual usable source. The principles of indexing and metadata introduced in this chapter will appear immediately and concretely in catalog searching.