There are two kinds of definitions relevant to data warehouse design - (1) definition of business terms (2) definition of data elements.
Data element definitions help the business person make sense of the data and to understand the meaning of data stored in their operational OLTP system. Business term definitions help the business person manage his business. For eg. a business term definition may be that a Premier customer is defined as a customer with a high credit rating. Now the data element definition may be "A premier customer is represented in the data warehouse as a type code P in the customer entity.
Dealing with discrepancies between past and present definitions
Definitions should reflect both what is in the systems from the past (the way the data was and is now) and the way the enterprise really does business now.
Suppose a company sells a product in units of Kgs and each Kg contain 3 bottles. Now one day the Marketing manager announces that because one of its competitors sells it product in six bottle Kgs, it should also do the same and the company changed the Ka volume to 6 bottles and adjusted the price accordingly. Now several years passed and th company desires to build a data warehouse. Now while building the data warehouse n distinction was made between the 3 & 6 bottle kgs. Then one day when an analysis w made about the number of products sold over a period of time, they noticed a significant glitch in the number of product sold ie. the company has sold only half of what they used The reason was that the 3 & 6 bottle Kgs were treated semantically same in the data warehouse.