Individual records of a data set include data and semantic information to describe the data. The records in the data set are encoded to generate an encoded data set using a compression function which is determined from semantic information that is common to multiple records of the data set. Multiple records of the encoded data set include the data without the common semantic information. The encoded data set is transmitted, or otherwise communicated, to a destination device along with an expansion function that includes the semantic information that is common to the multiple records of the data set. The destination device expands the encoded data set with the expansion function such that the multiple records of the encoded data set are expanded to include the common semantic information.