Geocoding method using multidimensional vector spaces

• US 7,627,545 B2
• Filed: 12/27/2005
• Issued: 12/01/2009
• Est. Priority Date: 02/02/2005
• Status: Expired due to Fees

• Current Assignee: AT&T Corporation (AT&T, Inc.)
• Original Assignee: AT&T Intellectual Property II LP (AT&T, Inc.)
• Inventors: Asher, Michael, Giddens, Charles, Eslambolchi, Hossein, Stewart, Harold J.
• Primary Examiner(s): Vincent, David R
• Assistant Examiner(s): Bharadwaj, Kalpana
• Application Number:
  US11/319,122
• Publication Number:
  US 20060245572A1
• Time in Patent Office: 1,435 Days
• Field of Search: None
• US Class Current: 706/52
• CPC Class Codes: G06F 16/29 Geographical information databases

Abstract

A process for evaluating and geocoding of GIS data elements utilizes a plurality of “locate” tests and a weighting scheme to express the match results as a multidimensional vector. Multiple inputs and data sources, as well as ambiguous and partial input data, are used to generate an output with improved precision by applying a weighting function to each input element and generating a set of test vectors (i.e., the input data element weighted by the known accuracy of the element/source). A sum of a plurality of tests is then generated as the “characteristic vector” of the test set. By using two (or more) different sets of test, two (or more) characteristic vectors are formed. Various well-known algebraic techniques can then be used to evaluate the results of each set of tests and select the “best match” result.

Claims

1. A method of evaluating a reliability of using ambiguous input geographic data to generate a geocoded output address match, the method comprising:

selecting a first plurality of n different types of locate tests to be used to convert the ambiguous input geographic data into the output address match;

retrieving a known certainty value (1−

u) associated with each test in the selected first plurality of n different types of locate tests;

defining a weighting constant (w) for each test in said selected first plurality of n different types of locate tests, the weighting constant within the range of 0.0 to 1.0; and

generating a characteristic vector defining the reliability of the output address match with respect to the ambiguous input geographic data, the characteristic vector defined as the sum of each known certainty value multiplied by its associated weighting constant for the first plurality of n different types of locate tests.

2. The method of evaluating reliability as defined in claim 1, wherein selecting is repeated for at least a second plurality of locate tests, using at least one different test than included in the first plurality of locate tests, to generate at least a second characteristic vector;

the method further comprising;

comparing the first and second characteristic vectors and defining a larger vector as being associated with a more reliable output address match.

3. The method of evaluating reliability as defined in claim 2 wherein the method is repeated for a multiple of N pluralities of different types of locate tests to generate a multiple of N characteristic vectors for comparison in evaluating reliability.

4. A computer-readable medium having embedded thereon a program, the program being executable by a processor to perform a method of evaluating a reliability of using ambiguous input geographic data to generate a geocoded output address match, the method comprising:

selecting a first plurality of n different types of locate tests to be used to convert the ambiguous input geographic data into the output address match;

retrieving a known certainty value (1−

u) associated with each test in the selected first plurality of n different types of locate tests;

defining a weighting constant (w) for each test in said selected first plurality of n different types of locate tests, the weighting constant within the range of 0.0 to 1.0; and

generating a characteristic vector defining the reliability of the output address match with respect to the ambiguous input geographic data, the characteristic vector defined as the sum of each known certainty value multiplied by its associated weighting constant for the first plurality of n different types of locate tests.

5. A computer-readable medium as defined in claim 4 wherein the embedded program is further executable to:

compare the first and second characteristic vectors and define a larger vector as being associated with a more reliable output address match.