Information Retrieval and Database Concepts in Biomedicine

A complete set of vocabulary flashcards covering Information Retrieval (IR) concepts, controlled terminologies, MeSH, DCMI, term weighting, and database architecture models based on the lecture transcript.

Information Retrieval (IR)

The acquisition, organization, and searching of knowledge-based information derived from observational or experimental research.

Primary Knowledge-Based Information

Original research literature that appears in scientific journals, books, reports, and other primary sources.

Secondary Knowledge-Based Information

Literature that reviews, condenses, or synthesizes primary literature, including textbooks, monographs, and review articles.

Open Access Publishing

A publishing model where authors or their institutions pay production costs upfront after peer review, making the paper freely available on the Web upon publication.

Bibliographic Content

Information resources that contain citations and references to biomedical literature, such as MEDLINE, which indexes around $5,000$ scientific journals.

Aggregated Content

Information compiled from multiple content types for target audiences such as consumers or clinicians, exemplified by MedlinePlus.

Indexing

The process of assigning metadata to content to facilitate its retrieval.

Manual Indexing

The assignment of indexing terms and attributes to documents by human indexers, often using a controlled terminology and specific protocol.

Automated Indexing

Computer-driven indexing that typically defines all consecutive alphanumeric sequences between white space as distinct word indexing terms.

Controlled Terminology

A standardized set of terms that can be applied to information processing tasks such as indexing.

Vocabulary

A controlled terminology that explicitly defines its terms.

Thesaurus

A controlled terminology that contains term variants or synonyms and establishes hierarchical, synonym, or related relationships between terms.

Medical Subject Headings (MeSH)

The NLM controlled terminology containing over $26,000$ subject headings organized hierarchically into $16$ trees, used to manually index biomedical databases.

Dublin Core Metadata Initiative (DCMI)

An organization that specified a standard $15$-element metadata schema (e.g., DC.title, DC.creator, DC.subject) for Web resource creation and indexing.

Inverse Document Frequency (IDF)

A measure used in term weighting, calculated as $IDF(\text{term}) = \log \frac{\text{number of documents in database}}{\text{number of documents with term}} + 1$.

Term Frequency (TF)

A measure of term occurrences within a given document, defined as $TF(\text{term, document}) = \text{frequency of term in document}$.

Synonymy

An automated indexing limitation where different words or phrases express the exact same concept (e.g., "hypertension" and "high blood pressure").

Polysemy

An automated indexing limitation where a single word possesses multiple meanings depending on context (e.g., "lead" as a chemical element vs. an ECG lead).

Granularity

An automated indexing limitation occurring when queries and documents describe concepts at different levels within a hierarchical structure.

PubMed

The retrieval system hosted by the National Library of Medicine (NLM) to search MEDLINE and other biomedical bibliographic databases.

Database

A structured, indexed repository of information that enables entering, extraction, cross-referencing, and analysis of data.

Record (Entry)

A horizontal row in a database table representing a single complete entity or entry.

Field

A vertical column in a database table representing a specific category or attribute that holds a value.

Flat File Database

A single long text file where entries are separated by special characters (such as |) and fields by commas or tabs, requiring full file scans to search.

Relational Database

A database structure storing data across multiple tables linked to one another by a shared unique key field in each record.

Object-Oriented Database

A database structure where data is stored as objects combining data and routines, linked by pointers to represent complex hierarchical relationships.

Information Retrieval Architecture

The interactive framework connecting Queries and Content to Metadata through Retrieval and Indexing operations managed by a Search Engine.

Relational Database Table Linkage

The connection of separate database tables (such as protab1 and protab2) using a unique shared key field (such as Protein-code).