Predicting Candidates Using Information Sources

• US 20090150322A1
• Filed: 12/07/2007
• Published: 06/11/2009
• Est. Priority Date: 12/07/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Morimoto, Kenichi, Davis, Peter, Bower, Jason, Robbins, Daryn Elliot, Laughlin, Chet
• Granted Patent: US 7,953,692 B2
• Time in Patent Office: Days
• US Class Current: 706/50
• CPC Class Codes: G06F 40/274 Converting codes to words; Guess-ahead of partial word inputs

Abstract

Embodiments are configured to provide one or more candidates based in part on an input. In an embodiment, a system includes a prediction engine which can be configured to provide one or more ranked candidates using one or more configurable data sources and/or candidate providers. Each data source can be configured to include a candidate provider to predict and return predicted candidates. The prediction engine can use a predicted candidate to rank and return a ranked candidate to a user interface, process, or other application. In one embodiment, a computing device can include and use a prediction engine which can use a customized ranking component to rank and return ranked candidates based in part on a query type. The customized ranking component can use predicted candidates provided by one or more data sources and/or candidate providers when making a ranking determination.

Claims

1. A system for providing information based on an input, the system comprising:

a plurality of data sources including;

a plurality of data stores, wherein each data store includes stored lexical data and other information; and

,

a plurality of candidate providers to provide one or more candidates and prediction information associated with the one or more provided candidates based in part on the input and the stored data of the data store; and

,

a ranking component to rank the one or more provided candidates, wherein the ranking component includes a ranking algorithm to output a ranked candidate based in part on the one or more provided candidates, the prediction information, and one or more confidence parameters associated with one or more of the plurality of data sources.

2. The system of claim 1, further comprising a managing component to direct the input to at least one of the plurality of data sources based in part on an input type associated with the input.

3. The system of claim 2, wherein the managing component is further configured to communicate the ranked candidate from the ranking component to a user interface (UI) used to provide the input.

4. The system of claim 2, wherein the managing component is further configured to communicate the ranked candidate to an application associated with the input.

5. The system of claim 1, further comprising a managing component to direct the input to at least one of the plurality of data sources based in part on a handwritten input.

6. The system of claim 1, further comprising a managing component to direct the input to at least one of the plurality of data sources based in part on a speech-based input.

7. The system of claim 1, further comprising a managing component to direct the input to at least one of the plurality of data sources based in part on a typed input.

8. The system of claim 1, wherein the ranking component can be customized by replacing the ranking algorithm with a custom ranking function configured to output the ranked candidate.

9. The system of claim 1, further comprising a managing component configured to forward a notification based on the input to the ranking component and at least one of the data sources, wherein the notification can be used to adjust a prediction configuration, wherein the prediction configuration is selected from a group consisting of a language model, a tailored ranking algorithm, and candidate provider.

10. The system of claim 1, wherein the plurality of data sources is available to disparate input methods that use lexicons for input.

11. The system of claim 1, wherein the ranking component include a ranking algorithm which comprises:

P(w)=s1×

f₁(w)+s2×

f₂(w)+s3×

f₃(w)+ . . . s(n)×

f_n(w)

wherein,

P(w) is an output which corresponds to the probability of a given candidate;

s1 . . . s(n) correspond to weights of the data sources; and

,

f1(w) . . . f_n(w) correspond to probabilities of candidates provided by the candidate providers.

12. The system of claim 11, wherein the ranking component is further configured to sort outputs of the ranking algorithm in a numerical order.

13. The system of claim 11, wherein the ranking component is further configured to provide a number of sorted candidate probabilities according to a specified number of ranked candidates.

14. A computer-readable storage medium including executable instructions which, when executed, predict information by:

receiving a query input, wherein the query input includes an input type;

predicting candidates based in part on the query input including directing the query input to a plurality of data sources based in part on the input type, wherein the plurality of data sources provide the predicted candidates;

ranking the predicted candidates based in part on a candidate probability and a data source weight to provide ranked candidates; and

,

ordering the ranked candidates in a numerical order.

15. The computer-readable storage medium of claim 14, further predicting information by ranking the predicted candidates with a custom ranking component.

16. The computer-readable storage medium of claim 14, further predicting information by predicting candidates using a number of custom data sources.

17. The computer-readable storage medium of claim 14, further predicting information by using an input type comprising a handwritten input, spoken input, or typed input.

18. A computer-implemented method of providing information comprising:

receiving an input associated with a user query;

predicting words based in part on the input using a plurality of data sources, wherein the plurality of data sources includes a plurality of providers having disparate prediction functionalities;

ranking the predicted words using a ranking algorithm to output a rank of a predicted word based on a predicted word probability and a confidence parameter to provide ranked words; and

,

sorting the ranked words.

19. The computer-implemented method of claim 18, further comprising ranking the predicted words using a ranking algorithm that outputs a rank of a predicted word based on a predicted word probability and a weight associated with one of the plurality of data sources.

20. The computer-implemented method of claim 18, further comprising replacing the ranking algorithm with a custom ranking algorithm to rank the candidates.