One of the most overlooked aspects of any information management project is metadata. When creating an electronic content management system, sending records to offsite storage or organizing a file room an organization’s metadata will facilitate the access and retrieval of information. Metadata organizes and identifies information; it can come from within content as well as externally. Knowledge management and metadata are inextricably linked as it will tell you the common language in your organization.
One way that metadata functions is through a thesaurus. The benefit of a thesaurus is it tracks how information changes over time with “use for” and “use instead” terms so you can see what other names people use for information. The narrower terms and broader terms in a thesaurus allow you to see the relationship of concepts in an information architecture. When information is touched by different content sets or departments, a thesaurus allows one to see the correlation with how information is used.
Using a semantic network allows one to get a birds-eye view of metadata at work. A semantic network also shows the relationship to terms as they connect to a certain concept. For example, a semantic network can show that a tire, engine and transmission are all parts of a car. A thesaurus is rigid in its structure as it describes how terms relate to one another. A semantic network provides more flexibility since it looks at concepts from a higher level. All thesauri will look the same and conceptually function the same, but semantic networks can take different shapes and forms based the concept they are representing.
If a thesaurus is a road map with street names such as “use for,” “use instead,” “narrow term,” etc., and a semantic network is like traveling by plane where you can go anywhere in any such way then a relational knowledge representation is like a game of 20 questions to figure out where you are going. A relational knowledge representation is presenting comparison often accomplished with a database. In a relational knowledge representation facts about a set of objects are systematically stored in columns to serve as the knowledge basis for an inference engine.
The information relationship building accomplished by thesauri, semantic networks and relational knowledge representation often serve as the backbone auto-classification software, auto categorization and entity extraction.