Geocoding using information retrieval

• US 8,626,789 B2
• Filed: 06/01/2007
• Issued: 01/07/2014
• Est. Priority Date: 06/01/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Jayanti, Harish, Shinn, Eugene
• Primary Examiner(s): Ehichioya, Fred I
• Assistant Examiner(s): OWYANG, MICHELLE N
• Application Number:
  US11/756,775
• Publication Number:
  US 20080301092A1
• Time in Patent Office: 2,412 Days
• Field of Search: 707705-780
• US Class Current: 707/771
• CPC Class Codes: G06F 16/29 Geographical information databases
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/9537 Spatial or temporal dependent retrieval, e.g. spatiotemporal queries

Abstract

Geocoding architecture based on information retrieval. The geocoding information retrieval system operates on locations and geographic entities. The architecture uses parameters in addition to the query string, if available. The parameters provide contextual information that enables filtering and sorting the results. The additional parameters include market, culture, map view, and user location. A geocoding triage process uses the query and context information provided and output of a machine-learning parser to select the data-lookup geocoders for processing the query. A results processing component collects the results from the underlying data-lookup geocoders, merges the results into a single list after normalizing the scores, and then re-sorts the results list.

Claims

1. A computer-implemented geocoding system, comprising:

a processor;

a triage component executing on the processor and configured to perform steps including;

receiving a query comprising a string and user context information, wherein the user context information includes at least one of market data, culture data, map view data and user location data, wherein the market data indicates a preferred geographical region, the culture data indicates a language setting, the map view data provides a geographic context of a current user view, and the user location data is based on a physical location of user;

parsing the string into a set of data-retrieval requests by extracting geographical information to generate a list of ranked geographical information using a weighting function and validating an address from the geographical information, wherein each data retrieval request comprises user context information and fields of a valid address, and wherein the set of data-retrieval requests comprises more than one data-retrieval request for execution by geocoders;

selecting a plurality of geocoders to which to send the set of data retrieval requests for execution based on specific requirements of each geocoder, wherein the specific requirements of a first geocoder are different from the specific requirements of a second geocoder, and the data retrieval requests sent to the first geocoder are different from the data retrieval requests send to the second geocoder; and

using at least one data retrieval request in the set of data retrieval requests to generate a geocoding command specifically for each of the selected plurality of geocoders to consume, wherein the geocoding command of the first geocoder is different from the geocoding command of the second geocoder;

a geocoding component executing on the processor and configured to perform steps including;

aggregating the set of data-retrieval requests; and

federating the set of data-retrieval requests and generated geocoding commands to the selected plurality of geocoders for parallel execution, wherein the selected plurality of geocoders return responses, wherein the responses comprise a set of results; and

a results component executing on the processor and configured to perform steps including;

processing the responses from the selected geocoders by merging the responses into a ranked list of results using predetermined scoring and ranking rules, and a subset of the results is returned.

2. The system of claim 1, wherein the string is part of a query of natural language text.

3. The system of claim 1, further comprising a query component executing on the processor and configured to perform steps including receiving the string in a query and adding context information to the string.

4. The system of claim 1, wherein the triage component executing on the processor and further configured to perform steps including performing a classification process on the string to obtain the more than one parse result.

5. The system of claim 4, wherein the classification process is based on machine learning.

6. The system of claim 1, wherein the results component executing on the processor and further configured to perform steps including generating scores for the results, and the results are ranked based on the scores.

7. The system of claim 1, wherein the results component executing on the processor and further configured to perform steps including normalizing static data-retrieval requests, and generates dynamic scores for the results based on multiple dimensions of relevance using at least one of context information, data-retrieval possibilities, the static data-retrieval requests, or user personalization.

8. A computer-implemented method of geocoding based on data retrieval, comprising:

receiving a query string accompanied by user context information, wherein the user context information includes at least one of;

market data, culture data, map view data or user location data, wherein the market data indicates a preferred geographical region, the culture data indicates a language setting, the map view data provides a geographic context of a current user view, and the user location data is based on a physical location of user;

parsing the query string into a set of data-retrieval requests by extracting geographical information to generate a list of ranked geographical information using a weighting function and validating an address from the geographical information, wherein each data retrieval request comprises user context information and fields of a valid address, and wherein the set of data-retrieval requests comprises more than one data-retrieval request for execution by geocoders;

selecting a plurality of geocoders to which to send the set of data retrieval requests for execution based on specific requirements of each geocoder, wherein the specific requirements of a first geocoder are different from the specific requirements of a second geocoder, and the data retrieval requests sent to the first geocoder are different from the data retrieval requests send to the second geocoder; and

using at least one data retrieval request in the set of data retrieval requests to generate a geocoding command specifically for each of the selected plurality of geocoders to consume, wherein the geocoding command of the first geocoder is different from the geocoding command of the second geocoder;

aggregating the set of data-retrieval requests;

federating the set of data-retrieval requests and generated geocoding commands to the selected plurality of data geocoders for parallel execution, wherein the selected plurality of geocoders return responses, wherein the responses comprise a set of results;

merging the responses from the selected data geocoders into a ranked list of results using predetermined scoring and ranking rules; and

selecting a set of results from the ranked list of results and returning the set of results for processing.

9. The method of claim 8, wherein the query string is a structured string or an unstructured natural language string.

10. The method of claim 8, further comprising:

applying scoring rules to results in the responses to generate scores;

applying weights to the scores;

aggregating the scores into a composites score; and

assigning the composite score to a result that spawned the composite score.

11. The method of claim 8, further comprising optimizing the query string based on country and region.

12. The method of claim 8, further comprising ranking the results based on a static rank and a dynamic rank.

13. A computer-implemented system, comprising:

a processor;

computer-implemented means for receiving a query comprising a natural language string of text from a client;

computer-implemented means for adding user context information to the query, wherein the user context information includes at least one of;

market data, culture data, map view data or user location data, wherein the market data indicates a preferred geographical region, the culture data indicates a language setting, the map view data provides a geographic context of a current user view, and the user location data is based on a physical location of user;

computer-implemented means for parsing the natural language string by the processor into more than one parse result;

computer-implemented means for creating a set of data-retrieval requests by extracting geographical information to generate a list of ranked geographical information using a weighting function and validating an address from the geographical information based on the more than one parse result, wherein each data retrieval request comprises the user context information and fields of a valid address, and wherein the set of data-retrieval requests comprising more than one data retrieval request for execution by geocoders;

computer-implemented means for selecting a plurality of geocoders to which to send the set of data retrieval requests for execution based on specific requirements of each geocoder, wherein the specific requirements of a first geocoder are different from the specific requirements of a second geocoder, and the data retrieval requests sent to the first geocoder are different from the data retrieval requests send to the second geocoder;

computer-implemented means for using at least one data retrieval request in the set of data retrieval requests to generate a geocoding command specifically for each of the selected plurality of geocoders to consume, wherein the geocoding command of the first geocoder is different from the geocoding command of the second geocoder;

computer-implemented means for federating the set of data-retrieval requests and generated geocoding commands to the selected plurality of data geocoders for parallel execution, wherein the selected plurality of geocoders return responses, wherein the responses comprise a set of results;

computer-implemented means for merging responses from the selected data geocoders into a ranked list of results using predetermined scoring and ranking rules; and

computer-implemented means for selecting a set of results from the ranked list of results and returning the set of results to the client.