 Metadata and documentation play an important role in RDM, enabling data to be found and re-used. Metadata is often defined as data that describes other data. If you have seen our knowledge clip about documentation, you might remember that the key difference between them, is that metadata records essential information about data in a highly structured way, using a set of defined information fields or elements. The reason why metadata is highly structured, is because it is meant to be readable and exchangeable by computers, something often referred to as machine readability. Metadata is needed for many things. It facilitates the process of searching and finding data. Metadata can help us to assess whether the data we find is useful for us or not, without having to download it first. It also lets us know how the data can be accessed, and how can it be re-used. Because of this, metadata is essential to make your data fair. Let's now have a look at some metadata concepts to understand why it is so important. First of all, there are different types of metadata. A first type is called descriptive metadata. This type includes common elements or fields that help us to discover the data. This can be for instance things like title of the dataset, the author, keywords describing the subject, and so on. When we talk about technical metadata, we mean information about technical aspects of the data or files. This could be for instance information about how to access the data, the file type used or the size of the file. Administrative metadata contains elements or fields that deal with intellectual property rights such as the license, or access rights or restrictions. Finally, there is also structural metadata. This type of metadata indicates how the dataset relates to other online resources. So, how is metadata created and where can we find it? Metadata can be associated to many different research objects and appear in many different ways. Sometimes, metadata is generated automatically. Some instruments such as microscopes, telescopes, or digital cameras create metadata when data is collected. But this is not always the case. Other times, metadata needs to be manually created. For instance by taking notes in a laboratory notebook, or by filling out a form or data listing. The second question is, how is metadata stored? Metadata can be stored embedded within the files, or it can be stored as separate files. And another way to provide metadata comes when you upload your data to a data repository or archive. Let's have a look at some details and examples. Most day-to-day digital files include a range of metadata fields. These allow you, for example, to search and sort files according to date created, file type, author, size, etc. Often, discipline-specific file formats might also have additional embedded metadata fields. For example, microscopy images normally include the objective settings within the file. Besides research instrumentation, metadata can also be generated by processing or analysis software. For example, statistical packages such as SPSS embed rich metadata within the file, like formats or additional variable information. It is important to find out whether the file formats you use of metadata fields embedded, and if these are needed to use the data. If you plan to convert a file with embedded metadata to a different file format, you should check whether these metadata will also be present in the new format. In some domains, another place where metadata can be found is in the header of the files. Typically, this is a section at the top of the document, preceding the data, containing a summary of the data, or information about the instrumentation settings, about the variables, etc. Often, this metadata header follows agreed conventions or standards, and the information it contains can be read by applications, processing software or algorithms. In other cases, a metadata header can be manually created by a researcher. For example, to provide contextual details about an interview in the transcription file. When metadata is generated by research instrumentation or software, it might also be stored on a separate file. For example, sensors and measurement devices often provide configuration or calibration files. And software used to process geographical data might store geospatial metadata, such as the coordinate system, in separate files. But these separate files can also be manually generated by the researcher. For example, in a readme file or a spreadsheet. Recording metadata in such a way can also be done in a structured way, and often templates are available to help you. Using readme files can be a useful way to collect metadata during the course of the project. However, this approach has some downsides. For example, there is a risk that the link between metadata and the data they represent is lost, for example when files are moved. Keeping some kind of metadata is certainly better than collecting no metadata at all. But, as a general rule, custom-made approaches make difficult for metadata to be machine-readable, and your data become less findable and reusable. And this takes us to our last point. Providing metadata on a data repository or archive. Depositing your data on a repository might be required by your institution or research funder policies, or by the journal in which you want to publish your results. Even if not required, it is a good research practice and will increase the fairness of your data, because data repositories provide functionalities to make your data more fair, including services to create and manage metadata. To upload your data to a repository, you will be required to fill in a user-friendly form to describe your data. All the fields in this form are in fact metadata fields, pre-configured to meet a specific metadata standard, allowing the result to become machine-readable. Then, what are metadata standards? When the information fields captured within a specific metadata set become widely used and accepted, it often evolves into a metadata standard. To put it simply, a metadata standard or metadata schema defines the set of elements that can or must be used to describe a resource. The standard also tells you how these elements should be named. And also, which values are allowed, or what the required format is for each of the elements. Some metadata standards are designed to be used across different scientific domains. Examples of such generic standards are the Dublin Core Standard, or Data Site. But there are also discipline-specific standards, which typically contain additional elements, to satisfy the needs of a particular scientific domain. For example, the ecological metadata language is used in ecology research and has additional elements, such as taxonomic coverage, to indicate which species are included in the dataset. Another example of a specialized metadata standard is the Data Documentation Initiative, or DDI. This standard contains elements, such as questionnaire specification, for research that involves surveys. The use of metadata standards facilitates data exchange by different systems or applications. In other words, it makes research metadata interoperable, on of the fair data principles. To recap, during the research process, metadata can be created in different ways and appear in multiple forms. An important use of metadata is to make your data findable and let others know how they can access it, and reuse it. Data repositories provide you the functionalities to create and manage machine-readable metadata, and therefore, make your data more fair. That is why it is a good idea to familiarize yourself with the kind of metadata that repositories require. During the course of the project, make sure to document this information, so that when the time comes to provide the metadata, you are not only relying on your fading memory. For more information about metadata and data repositories, have a look at our website.