Interpreting Query Results

What is the Difference Between a Concatenated Data Format and a Related Data Format?

Archaeological data are hierarchical. Consider a ceramic sherd. The sherd has properties of it own: a single ware type, a single maximum thickness, and a single weight. But that sherd also may carry multiple decorative elements that are executed in a different design, color, and technique. Or consider the fill of a post hole. That fill has one length, width and depth. But it may also have multiple sediment inclusions whose size and frequency are important.

These are examples of one-to-many relationships. One sherd may have many decorative attributes. Posthole fill may have multiple sediment inclusions. Archaeological data, both artifactual and contextual, are filled with these one-to-many relationships. Relational databases were invented to handle just such hierarchical relationships among objects. They do so by recording variable values for objects at each level in the hierarchy in a separate table and by identifying which constituent objects belong to which higher-level objects by means of a unique identifier or “key”.

CMAP data structures take full advantage of the relational database model. This means that when data on both objects and their lower-level constituents are joined and extracted for analysis, the resulting data file may contain multiple records for a single object. Together, the multiple records contain the information on the constituents. However, they duplicate information on the primary object. The Artifact ID number is a good way to see whether the multiple records represent a single artifact.

This example shows how CMAP returns multiple records for a single artifact. Note the differences in Decorative Technique + Dech Tech Color:

This result may present difficulties for users who traditionally expect a one-to-one relationship between observations or records in a data file and higher-level objects (e.g. one sherd generates one record). For other users, preserving patterns of covariation among variables that characterize lower level constituents is analytically important, and this requires multiple records per object.

We have thus provided users with two different options for viewing and downloading data: a Concatenated Data Format and a Related Data Format. If users choose to download data in the Related Data Format, multiple records will be returned for every artifact that has more than one constituent object or part. Below is an example showing two different glass sherds. Note that there is more than one record for each artifact since each unique glass bottle constituent body element is being recorded (note the different artifact ids, 1000-109 and 1000-110).

Users who select to view and download data in the Concatenated Data Format will receive data sets that do not contain multiple records for a single record. Below is an example of the same artifacts returned in Concatenated Data Format. One record is returned for each artifact and the unique attributes are concatenated into one field, with each term separated by a comma and each record separated by a semi-colon.