Discovering schema using anchor attributes

A system for partitioning a data set, the system comprising:

• an electronic processor configured to:
• receive a plurality of initial anchor attributes of the data set, wherein the data set comprises a plurality of records, wherein each record comprises a plurality of attribute-value pairs, and wherein each initial anchor attribute defines a schema of records;
• determine a set of filter attributes from the attributes of the attribute-value pairs that commonly occur with the initial anchor attributes that are mutually exclusive with any of the plurality of anchor attributes;
• determine a set of candidate anchor attributes from the attributes of the attribute-value pairs comprising each unique attribute from the data set excluding the plurality of anchor attributes and the set of filter attributes;
• determine, for each of the candidate anchor attributes and the plurality of anchor attributes, an attribute context;
• determine, for each of the candidate anchor attributes, a context similarity between each of the plurality of anchor attributes;
• and
• determine a new anchor attribute from the set of candidate anchor attributes based on the context similarity, wherein the new anchor attribute defines a new schema of records.

A computer-readable storage media storing computer-executable instructions for partitioning a data set, the stored instructions comprising:

• instructions to receive a plurality of initial anchor attributes of the data set, wherein the data set comprises a plurality of records, wherein each record comprises a plurality of attribute-value pairs, and wherein each anchor attribute defines a schema of records;
• instructions to determine a set of filter attributes from the attributes of the attribute-value pairs that commonly occur with the initial anchor attributes and that are not mutually exclusive with any of the plurality of anchor attributes;
• instructions to determine a set of candidate anchor attributes from the attributes of the attribute-value pairs comprising each unique attribute from the data set excluding the plurality of anchor attributes and the set of filter attributes;
• instructions to determine, for each of the candidate anchor attributes and the plurality of anchor attributes, an attribute context;
• instructions to determine, for each of the candidate anchor attributes, a context similarity between each of the plurality of anchor attributes;
• and
• instructions to determine a new anchor attribute from the set of candidate anchor attributes based on the context similarity, wherein the new anchor attribute defines a new schema of records.