The simplest description of a Semantic Database is that it is a database of sentences. Every little nugget of data: the field of a record, the cell of a table, the branch of a tree, the relationship between objects, can all be represented by a sentence. So, if you have a database of sentences, you can hold any data.
Sentences are composed of a Subject, a Verb, and an Object (SVO), which are called "semantic triples":
The Subject is the RowID or ObjectID.
The Verb is the ColumnName or FieldName.
The Object is the FieldValue or CellValue.
Semantic Triples representing Object Tree
Datura invented an advanced version of the semantic database, called the semantic-relational database. The additions that were added were:
storage in a relational database, indexed and accessible with SQL
different data types, indexable by type (such as numbers and date-time values)
global, multi-part IDs to make it easier to share data and update sentences
option to expand sentences with metadata (such as units)
optional sentences-about-sentences, tagging sentences with metadata
Datura used this patented technology to build the Legume Data System, which forms the core of its data products.