Metadata management for unstructured data sources poses a major challenge due to the need to capture the metadata. Unlike structured data sources, which have predefined elements and values, unstructured data sources require manual or automated methods. Parsing is a common technique, where natural language processing (NLP) or other methods are used to analyze the text or content of the data source and identify key metadata elements, such as author, date, topic, sentiment, and keywords. Tagging is another method which involves assigning labels or categories to the data source based on its content, context, or purpose. This can be done manually by creators or consumers, or automatically using machine learning or other algorithms. Lastly, enriching involves adding additional metadata elements or values to the data source based on external sources or rules. For example, one can enrich an email's metadata by adding the location, organization, or role of the sender/recipient, as well as by linking it to a project or customer.