Method and apparatus for constructing a new database from overlapping databases

A method for use in a data processing system for constructing a resultant database from a first database and a second database wherein data to appear in the resultant database is present in modified form in the first and second databases containing data records, the data processing system including a memory for storing the first, second and resulting databases, a processor for performing operations upon data records of the first, second and resulting databases and a plurality of procedures for controlling the operations of the processor, the method comprising the steps of:

• selecting from the first database at least one construction property to be used in constructing the resultant database
• wherein a construction property is a type of information represented in partly or wholly implicit form by values stored in the data records of the first database and in explicit form in the values stored second database and
• wherein there is partial congruence of the at least one construction property between the first, second and resultant databases,
• selecting from the first and second databases at least two modeling properties
• wherein a modeling property is a type of information represented by values stored in the data records of the first and second databases and
• wherein there is partial congruence of the at least two modeling properties between the first, second and resultant databases,
• by operation of the processor operating under control of an inclusion probability procedure, assigning an inclusion probability to each of the data records of the first database containing the selected construction and modeling properties,
• by operation of the processor operating under control of a modeling procedure,
• constructing the modeling properties of the data records of second database containing the selected modeling properties,
• modeling the modeling properties of the data records of the first database containing the selected modeling properties against the constructed modeling properties of the data records of the second database containing the selected modeling properties, and
• comparing the modeled modeling properties of the data records of the first database and the constructed modeling properties of the second database, and
• by operation of the processor operating under control of the inclusion probability procedure, adjusting the inclusion probability of each one of the data records of the first database containing the selected modeling properties in accordance with the comparison of the modeled modeling properties of the data records of the first database and the constructed modeling properties of the second database until an acceptable match is achieved between all the modeled modeling properties of the data records of the first and the constructed modeling properties of the second database, and
• by operation of the processor operating under control of a resultant data record construction procedure, constructing data records of the resultant database from the finally selected data records of the first database using a probabilistic assignment operation in accordance with the final inclusion probabilities assigned to the plurality of data records of the first database to select the data records of the first database to be used in construction of the resultant database.