メタデータとは何か？

What is Metadata? | IBM Tags Artificial Intelligence Analytics What is metadata? Authors Annie Badman Staff Writer IBM Think Matthew Kosinski Staff Editor IBM Think What is metadata? Metadata is information—such as author, creation date or file size—that describes a data point or data set. Metadata can improve a data system’s functions and make it easier to search for, organize and use data . The simple definition of metadata is "data about data." That is, it provides details about data that is separate from the content of the data itself. For example, a table of recent customer orders constitutes a data set, while information about that data set—such as who owns it or what type of file it is—is the metadata. Different types of metadata serve different functions. For example, descriptive metadata for a document might include the author, creation date, file size and keywords. Technical metadata for a relational database might describe the structure of tables, data types and relationships between tables. Given the sheer volume of data that businesses handle today, metadata is indispensable. It enhances data accessibility, making it easier to navigate large data sets and convert raw information into actionable insights. For example, a retail business can use metadata to quickly find sales data for a specific month, filtered by product category and region, without searching through all its data. Metadata is also critical in data governance and data management . In fact, Gartner found that enterprises that don’t take a metadata-driven approach to IT modernization can spend as much as 40% more on data management. 1 Systems such as databases , digital libraries and content management platforms rely on metadata to sort, retrieve and manage data. When data is better organized, organizations can maximize its value for critical business initiatives, including business intelligence (BI), artificial intelligence (AI) and machine learning (ML) projects. Metadata also helps ensure data quality and data integrity . It enables data lineage efforts and supports compliance with regulations, such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA) . Visualizing metadata Metadata can, at times, feel a little meta. To make it more concrete, consider a book. The metadata here would be the author, title, publication date and table of contents. These things don't provide the actual data—the book's text—but they do provide essential details for classifying the book and understanding its origins. Just as a librarian or researcher might parse thousands of titles by using an author's name, data scientists or other users can search large data sets by using metadata. Without this information, users would encounter massive amounts of data—akin to hundreds of thousands of uncategorized pages of text—with no ability to sort through or organize it. In other words, the data, such as the books, would lose all context and function. Stay ahead of the latest tech news Weekly insights, research and expert views on AI, security, cloud and more in the Think Newsletter. Subscribe today Types of metadata There are several types of metadata, which all describe different aspects of data. For instance, different types of metadata might describe the content of a computer file, its structure or its permissions. Some of the most common kinds of metadata include: Descriptive metadata Structural metadata Administrative metadata Technical metadata Preservation metadata Descriptive metadata Descriptive metadata provides basic information about data, such as file title, author, keywords and summary. Descriptive metadata is commonly used in social media platforms, search engines and data catalogs because it helps make data searchable. On LinkedIn, for example, descriptive metadata—such as a user’s job title, education and skills—can help sort and identify profiles. Semantic metadata, while technically more of a framework than a type of metadata, can complement descriptive metadata. Semantic metadata defines relationships between data points and adds contextual meaning. For example, if a data set contains the term "automobile," semantic metadata can provide clarity on how "automobile" relates to other terms such as "vehicle" or "SUV." Structural metadata Structural metadata defines how data elements are organized and related to each other. For example, on a website, structural metadata establishes how the home page links to other subpages and categorizes them into sections. Structural metadata also helps systems organize complex data, similar to how a table of contents helps guide readers through a book. Taxonomies—organized systems for categorizing data—often guide structural metadata. For example, a retailer might use a taxonomy to classify different products, putting "mobile phones" under "electronics" and “t-shirts” under “clothing.” Administrative metadata Administrative metadata provides information on data ownership, access permissions and retention policies. For example, it might detail who created the data, who can modify it and how long to store it. Administrative metadata can also help ensure compliance with legal and organizational rules by making a record of data access and managing retention schedules. Also, it plays a critical role in managing preservation metadata, helping ensure that data remains accessible and usable over time. Technical metadata Technical metadata describes a data file's technical details, such as file type, encoding information and storage location. For example, the technical metadata for an image might include its resolution, file size, file format and color profile. Technical metadata also helps ensure data is stored, processed and displayed correctly across various systems and platforms, such as content management systems or cloud storage solutions. For example, technical metadata includes crucial information that different systems and environments need to accurately interpret and present data, such as the proper resolution for displaying an image. Preservation metadata Preservation metadata helps ensure the long-term usability and accessibility of data. It includes details about a data set's last backup and strategies for preserving data, such as porting data to new formats as technology evolves to keep it from becoming obsolete. For example, a healthcare organization might use preservation metadata to convert patient records from legacy systems to modern electronic health record (EHR) formats for continued access and compliance. Preservation metadata is critical in industries such as healthcare and legal services, where organizations must retain data for extended periods to comply with certain rules and regulations. The role of metadata in data management Because metadata is crucial for organizing, searching and accessing data, it is critical to every step of the data management process, from data creation and storage to retrieval and archiving. Here is a breakdown of how metadata fits into each step of the data management lifecycle. Data creation Data creation is nearly always accompanied by metadata creation, either manually or automatically. For example, when someone takes a digital photo, most digital cameras automatically record metadata such as the camera model, date and time. Similarly, when users upload a data set to a data repository, they can also manually add descriptive metadata to improve searchability. Data storage and organization Metadata provides organizations with a structure to categorize, describe and organize large volumes of data. This enables organizations to store data in a more logical and coherent way. In relational databases , for example, metadata specifies how tables are connected, making analysis easier. Metadata can add labels, descriptions or tags to unstructured data to help systems sort files and records more efficiently. Metadata also supports version control by tracking modifications and file origins. Data access and retrieval Metadata makes stored data more searchable and retrievable. Instead of manually sifting through vast amounts of information, users can quickly locate specific data by using metadata such as keywords, file descriptions or creation dates. For example, metadata tags in a content management system can help users sort content by publication date, author or topic. Search engines depend on metadata embedded in HTML headers, such as title tags and descriptions, to rank and deliver relevant web pages. Data archiving and preservation As data reaches the end of its active use, metadata helps ensure proper archiving and long-term preservation. Preservation metadata documents important details such as file format, access permissions, last modification date and backup history. By tracking these details, organizations can help ensure that archived data remains usable for years or even decades while remaining compliant with regulatory standards. Metadata standards and schemas Metadata standards and schemas help ensure consistency across data sets, making it easier for systems and users to understand and share data. A metadata standard defines how metadata should be structured and applied across different systems. Two widely recognized metadata standards are Dublin Core and ISO 19115. Dublin Core is a simple and widely used standard for describing digital resources. It includes basic metadata elements such as "title," "author" and "format." This standard is often used in digital libraries and archives to help organize both physical and digital collections. ISO 19115 focuses on describing geospatial data. It helps ensure consistent documentation of geographic data sets, including details on data quality, geographic extent and methodology. This standard is critical for industries involved in mapping, land surveying and environmental monitoring. A metadata schema is a blueprint for implementing metadata standards. It helps ensure metadata elem