Back to results
US 8,943,057 B2Granted
Method and system for distributed bulk matching and loading
AssigneeOracle America, Inc.
InventorSwaranjit Singh Dua
Life of this patent
Estimating the term…
Abstract
A method for matching bulk data, involving receiving a plurality of input records, performing block distribution on the plurality of input records to obtain a plurality of block records, wherein performing block distribution comprises distributing the plurality of input records to a first plurality of containers, performing concurrent matching operations on the plurality of block records to obtain a plurality of single best records, and loading the plurality of single best records into a master index database.
Claims17
A method for matching bulk data, comprising:
- receiving a plurality of input records;
- performing block distribution on the plurality of input records to obtain a plurality of block records, wherein performing block distribution comprises distributing the plurality of input records to a first plurality of containers;
- performing concurrent matching operations on the plurality of block records to obtain a plurality of single best records,
- wherein performing concurrent matching operations on the plurality of block records comprises matching block records in each container of the first plurality of containers by:
- determining a match threshold weight;
- comparing a match weight for a pair of input records in one of the plurality of block records to the match threshold weight;
- and
- determining the pair of input records to be a match when the match weight of the pair of input records is not less than the match threshold weight,
- wherein the matched pair of input records is used to obtain one of the plurality of single best records, wherein each of the plurality of single best records comprises a unified profile of data compiled from a matched set of input records;
- and
- loading the plurality of single best records into a master index database.
8 more elements
A system for matching bulk data, comprising:
- a master computing device configured to:
- receive a plurality of input records, and
- perform block distribution on the plurality of input records to obtain a plurality of block records, wherein performing block distribution comprises distributing the plurality of input records to a first plurality of containers;
- a plurality of slave computing devices configured to execute concurrently to:
- match the plurality of block records to obtain a plurality of single best records by:
- matching block records in each container of the first plurality of containers to obtain a plurality of matched sets of input records by:
- determining a match threshold weight;
- comparing a match weight for a pair of input records in one of the plurality of block records to the match threshold weight;
- and
- determining the pair of input records to be a match when the match weight of the pair of input records is not less than the match threshold weight,
- wherein the matched pair of input records is used to obtain one of the plurality of single best records, wherein each of the plurality of single best records comprises a unified profile of data compiled from a matched set of input records;
- and
- a master index database configured to store the plurality of single best records.
10 more elements
A computer readable medium comprising instructions for matching bulk data, the instructions comprising functionality to:
- receive a plurality of input records;
- perform block distribution on the plurality of input records to obtain a plurality of block records, wherein performing block distribution comprises distributing the plurality of input records to a first plurality of containers;
- perform concurrent matching operations on the plurality of block records to obtain a plurality of single best records,
- wherein performing concurrent matching operations on the plurality of block records comprises matching block records in each container of the first plurality of containers by:
- determining a match threshold weight;
- comparing a match weight for a pair of input records in one of the plurality of block records to the match threshold weight;
- and
- determining the pair of input records to be a match when the match weight of the pair of input records is not less than the match threshold weight,
- wherein the matched pair of input records is used to obtain one of the plurality of single best records, wherein each of the plurality of single best records comprises a unified profile of data compiled from a matched set of input records;
- and
- load the plurality of single best records into a master index database.
8 more elements
Dependencies are inferred from the claim text.