File search glossary: 37 terms in plain English

Search tool documentation assumes you already know what an MFT, an iFilter or a partial hash is. This glossary fixes that: every term is defined in two to four sentences of plain English, with a link to the guide or review where it matters in practice. Each entry has its own anchor, so you can link straight to a definition. Newer to all of this? Start with our file search FAQ instead.

B

Boolean operators

Words like AND, OR and NOT that combine search terms: invoice AND 2026 requires both words, draft OR final accepts either, and NOT archived excludes matches. Support varies widely — Agent Ransack has the most complete boolean engine among free Windows tools, including nested expressions like (report OR summary) AND 2026.

C

Checksum

A short value computed from a file's bytes, used to verify integrity: if the checksum changes, the contents changed. Download pages publish checksums so you can confirm a file arrived intact, and duplicate finders rely on the cryptographic kind, better known as a hash.

Content search

Searching the text inside files rather than their names — the only way to find a document when you remember a phrase but not the filename. It is slower than filename search because every file must be opened and read, or indexed in advance. Our guide to searching file contents on Windows covers both approaches.

D

Duplicate detection

Finding files with identical contents, usually by grouping files by size first and then hashing the candidates. Reliable tools compare bytes, not filenames — two files called IMG_0042.jpg can be different photos, and identical photos can have different names. See our best duplicate file finders roundup for tools that get this right.

E

ETP server

Everything Transfer Protocol — a feature of Everything that publishes one machine's file index over the network so another Everything instance can search it instantly. It is the fastest way to search a Windows machine remotely, and a key trick in our network drive search guide.

exFAT

A filesystem designed for flash and external drives, readable by Windows, macOS and Linux alike — which is why it ships on most portable drives. It has no MFT or USN journal, so instant-search tools fall back to slower folder scans on exFAT volumes.

F

FAT32

A legacy filesystem still found on USB sticks and SD cards. Its hard 4 GB per-file limit and lack of journaling make it a poor choice for anything except small portable media that needs to work in old devices.

File attributes

Flags the filesystem stores for each file, such as hidden, read-only and system. Search tools can filter on them — Everything's attrib:H finds hidden files, for example — which is how you surface files Explorer normally hides from you.

File extension

The suffix after the last dot in a filename (.pdf, .docx, .tar.gz) that tells the operating system what type a file claims to be. Renaming the extension does not change the contents — see file signature for how the real type is identified.

File signature

The first few bytes of a file — the "magic bytes" — that identify its real format regardless of extension: %PDF opens every PDF, PK opens every zip-based file including .docx. Recovery and forensic tools identify files by signature, which is why they can recover files with no names at all.

Filename search

Matching only the names and paths of files, never their contents. Because names live in a compact structure like the MFT, filename search can be effectively instant across millions of files — it is the speed headline behind most tools in our best file search software ranking.

Fuzzy search

Matching that tolerates typos, abbreviations and out-of-order letters, so qrep still finds quarterly-report.xlsx. Launchers like Listary lean on fuzzy matching heavily, because when you are typing from memory you rarely type the exact filename.

G

Glob

A simple pattern style built from wildcards, such as *.log or report-??.docx. Globs are easier to write than regex and supported by almost every shell and search tool, but they can only express "anything here" — not "exactly four digits here".

H

Hash (MD5 / SHA-256)

A fixed-length fingerprint computed from a file's contents: if two files share a hash, their bytes are identical for all practical purposes. MD5 is fast and perfectly adequate for finding duplicate files; SHA-256 is the stronger choice when you need to rule out deliberate tampering.

I

iFilter

A Windows plugin that teaches the search indexer how to extract text from a specific file format. The classic example is the PDF iFilter, which lets Windows Search and Agent Ransack read inside PDFs — our PDF search guide shows how to check yours is working.

Index

A prebuilt database that maps filenames or words to their locations, so a search becomes a millisecond lookup instead of a minutes-long crawl. The trade-off is build time, disk space and freshness — an index is only as current as its last update.

Indexing

The process of building and maintaining that index: scanning files once up front, then watching for changes. Heavy disk activity during initial indexing is normal; constant re-indexing weeks later is a misconfiguration, and our guide to fixing slow Windows search shows how to diagnose it.

L

Launcher

A keyboard-summoned search bar for opening apps, files and actions — Listary, Alfred and Raycast are the best-known examples. Launchers optimize for getting the top hit right; dedicated search tools optimize for showing you the complete result list.

Lucene

An open-source full-text search engine library, originally Java, that powers many search products — including DocFetcher. It handles tokenizing, indexing and relevance ranking so individual tool authors don't have to reinvent them.

M

Mapped drive

A network share assigned a Windows drive letter such as Z:, making it look like a local disk. It is a convenience layer over a UNC path — search speed is still bound by the network and the server, not the letter.

mdfind

The macOS command-line interface to Spotlight: mdfind budget searches the Spotlight index instantly, and -onlyin ~/Documents limits it to one folder. Our Mac file search guide has a set of copy-paste examples.

Metadata

Data about a file rather than its contents: size, timestamps, author, camera EXIF data, tags. Because metadata is small and structured, searching it can filter millions of files without opening a single one — that is what powers filters like kind:image taken:2024.

MFT (Master File Table)

NTFS's master record of every file and folder on a volume — name, size, timestamps, location. Tools like Everything and UltraSearch read the MFT directly instead of walking folders, which is how they list millions of files in a few seconds.

N

NAS (network-attached storage)

A small always-on box of drives that serves files over your network to every device in the house. Most NAS systems can run their own search index, which beats crawling the share from a PC — our NAS buying guide covers models that do this well.

NTFS

The default Windows filesystem. Its MFT and USN journal are what make instant filename search and real-time index updates possible — features you lose the moment a drive is formatted exFAT or FAT32.

O

OCR (optical character recognition)

Software that converts images of text — scans, photos, faxes — into real, searchable text. A scanned PDF is invisible to every search tool until OCR adds a hidden text layer; our guide to searching inside PDFs walks through free and paid ways to run it.

P

Partial hash

Hashing only the first and last chunks of a file to triage duplicate candidates quickly, then confirming survivors with a full hash. On a photo library of 100,000 files this can cut scan time from hours to minutes, which is why the better tools in our duplicate finder roundup use it.

Portable app

A program that runs without installation — typically from a USB stick — and leaves no traces in the registry. Everything, Agent Ransack and DocFetcher all offer portable builds, which makes them handy on locked-down work machines; see our free tools roundup for which portable build to grab.

Q

Query operators

Prefix filters that narrow a search, such as ext:pdf, size:>1gb, dm:today or kind:document. Every serious tool has its own set, and learning five of them is the single biggest speed win in file search — our Windows search hub lists the ones built into Explorer.

R

Regex (regular expressions)

A pattern language for matching text with precision: invoice-\d{4} matches "invoice-" followed by exactly four digits, something no wildcard can express. The learning curve is real, so we keep ten copy-paste regex patterns that cover the most common file search jobs.

S

Search syntax

The mini-language a search tool understands — its operators, quoting rules and modifiers. Everything, Explorer and Spotlight each speak their own dialect, but the concepts transfer: once you have learned one, the next takes minutes.

SMB (Server Message Block)

The protocol behind Windows shared folders and most NAS shares. Searching over SMB is slow without a server-side index, because every file listing is a network round trip — our network drive search guide covers the workarounds.

Spotlight

macOS's built-in indexed search, covering filenames, contents and metadata, summoned with Cmd-Space. When its results go stale, rebuilding the index fixes most problems — the steps are in our Mac search guide, along with the operators Spotlight understands.

T

Timestamps (created / modified / accessed)

The three dates a filesystem keeps for every file. Modified is the most reliable for finding recent work; created can survive copies oddly, and accessed is often disabled or imprecise for performance reasons, so don't build a search on it.

U

UNC path

The direct address of a network resource, written \\server\share\folder. It works without mapping a drive letter, and most search tools accept UNC paths directly — useful when a mapped drive keeps disconnecting.

USN journal

NTFS's change log, which records every create, rename and delete on a volume. Tools like Everything subscribe to it to keep their index current in real time — no rescanning, no stale results, almost no CPU cost between searches.

W

Wildcard

Placeholder characters in a search pattern: * matches any run of characters and ? matches exactly one, so 20??-report.* finds every year's report in any format. Quick for everyday filters; switch to regex when you need real precision.