Topically aware word suggestions

• US 9,928,232 B2
• Filed: 02/10/2016
• Issued: 03/27/2018
• Est. Priority Date: 02/27/2015
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Rhodes, Alexander C.
• Primary Examiner(s): SPOONER, LAMONT M
• Application Number:
  US15/040,947
• Publication Number:
  US 20160253312A1
• Time in Patent Office: 776 Days
• Field of Search: 704 1, 704 9, 704 10
• CPC Class Codes: G06F 40/274 Converting codes to words; Guess-ahead of partial word inputs
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates

Abstract

Concepts and technologies are described herein for providing topically aware word suggestions. Using a text input, the system determines a conditional count and an unconditional count. The system then determines an adjustment factor for a pair of words of the plurality of words based on the unconditional count and the conditional count. The system then generates a data structure defining a plurality of word clusters. The system then reconstructs the adjustment factor of the pair of words based on a number of common clusters between individual words of the pair of words. The adjustment factor is combined with other data, such as data from a language model dictionary and a freshness factor from an average cluster activation state table to determine a probability associated with a word candidate, which is displayed to a user.

Claims

1. A method for generating a word candidate to assist a user providing an input to a computing device, comprising:

receiving, at the computing device, the input containing a plurality of words, wherein the computing device performs the operations of;

determining a conditional count;

determining an unconditional count;

determining an adjustment factor fora pair of words of the plurality of words based on the unconditional count and the conditional count;

generating a data structure defining a plurality of word clusters, individual word clusters of the plurality of word clusters include at least one word of the plurality of words;

reconstructing the adjustment factor of the pair of words based on a number of common clusters between individual words of the pair of words;

determining a candidate probability associated with the word candidate based, at least in part, on the reconstructed adjustment factor, wherein the word candidate is selected from individual words associated with the plurality of word clusters;

generating an output containing the word candidate based, at least in part, on the candidate probability; and

displaying the word candidate on a display screen of the computing device.

2. The method of claim 1, further comprising:

obtaining an input indicating a word; and

reconstructing a freshness value associated with one or more word clusters containing the word, the modification to the freshness value indicating that the one or more word clusters containing the word are more recent than other word clusters of the plurality of word clusters.

3. The method of claim 1, further comprising:

receiving a text entry;

determining one or more word clusters of the plurality of word clusters associated with the text entry;

obtaining a freshness factor associated with the one or more word clusters of the plurality of word clusters associated with the text entry;

obtaining a related adjustment factor associated with the one or more word clusters;

obtaining a language model value; and

determining the candidate probability associated with the word candidate based, at least in part, on the language model value and the related adjustment factor, wherein the word candidate is selected from individual words associated with the plurality of word clusters.

4. The method of claim 3, further comprising:

determining a plurality of word candidates, wherein individual words of the plurality of word candidates comprise an individual candidate probability based, at least in part, on the language model value and the related adjustment factor;

generating data indicating a ranking of the word candidate and the individual words of the plurality of word candidates based, at least in part, on the individual candidate probabilities and the candidate probability; and

generating an output indicating the ranking.

5. The method of claim 4, wherein the language model value comprises a probability in which a word associated with the text entry is universally used in a null context.

6. The method of claim 1, wherein reconstructing the adjustment factor of the pair of words is also based on a ranking of at least one correlation between the words.

7. The method of claim 1, wherein reconstructing the adjustment factor of the pair of words comprises:

determining a cluster density for individual clusters of the plurality of clusters;

determining an ordering of the plurality of clusters based on the cluster density for the individual clusters; and

reconstructing the adjustment factor based on the ordering of the plurality of clusters and the number of common clusters between individual words of the pair of words.

8. A computing device for generating a word candidate to assist a user providing an input to the computing device, comprising:

a processor; and

a memory having a set of computer-executable instructions stored thereupon which, when executed by the processor, cause the computing device to

receive, at the computing device, the input containing a plurality of words;

determine a conditional count;

determine an unconditional count;

determine an adjustment factor for a pair of words of the plurality of words based on the unconditional count and the conditional count;

generate a data structure defining a plurality of word clusters, individual word clusters of the plurality of word clusters include at least one word of the plurality of words;

reconstruct the adjustment factor of the pair of words based on a number of common clusters between individual words of the pair of words;

determine a candidate probability associated with the word candidate based, at least in part, on the reconstructed adjustment factor, wherein the word candidate is selected from individual words associated with the plurality of word clusters;

generate an output containing the word candidate based, at least in part, on the candidate probability; and

display the word candidate on a display screen of the computing device.

9. The computing device of claim 8, wherein the computer-executable instructions cause the computing device to:

obtain an input indicating a word; and

reconstruct a freshness value associated with one or more word clusters containing the word, the modification to the freshness value indicating that the one or more word clusters containing the word are more recent than other word clusters of the plurality of word clusters.

10. The computing device of claim 8, wherein the computer-executable instructions cause the computing device to:

receive a text entry;

determine one or more word clusters of the plurality of word clusters associated with the text entry;

obtain a freshness factor associated with the one or more word clusters of the plurality of word clusters associated with the text entry;

obtain a related adjustment factor associated with the one or more word clusters;

obtain a language model value; and

determine the candidate probability associated with the word candidate based, at least in part, on the language model value and the related adjustment factor, wherein the word candidate is selected from individual words associated with the plurality of word clusters.

11. The computing device of claim 10, wherein the computer-executable instructions cause the computing device to:

determine a plurality of word candidates, wherein individual words of the plurality of word candidates comprise an individual candidate probability based, at least in part, on the language model value and the related adjustment factor;

generate data indicating a rank of the word candidate and the individual words of the plurality of word candidates based, at least in part, on the individual candidate probabilities and the candidate probability; and

display data indicating the rank of the word candidate and the individual words.

12. The computing device of claim 11, wherein the language model value comprises a probability in which a word associated with the text entry is universally used in a null context.

13. The computing device of claim 8, wherein reconstructing the adjustment factor of the pair of words is also based on a ranking of at least one correlation between the words.

14. The computing device of claim 8, wherein reconstructing the adjustment factor of the pair of words comprises:

determining a cluster density for individual clusters of the plurality of clusters;

determining an ordering of the plurality of clusters based on the cluster density for the individual clusters; and

reconstructing the adjustment factor based on the ordering of the plurality of clusters and the number of common clusters between individual words of the pair of words.

15. A computer-readable storage medium having computer-executable instructions stored thereupon which, when executed by a computing device, cause the computing device to generate a word candidate to assist a user providing an input to the computing device, wherein the computing device performs the operations of:

receiving, at the computing device, the input containing a plurality of words;

determining a conditional count;

determining an unconditional count;

determining an adjustment factor for a pair of words of the plurality of words based on the unconditional count and the conditional count;

generating a data structure defining a plurality of word clusters, individual word clusters of the plurality of word clusters include at least one word of the plurality of words;

reconstructing the adjustment factor of the pair of words based on a number of common clusters between individual words of the pair of words;

determining a candidate probability associated with the word candidate based, at least in part, on the reconstructed adjustment factor, wherein the word candidate is selected from individual words associated with the plurality of word clusters;

generating an output containing the word candidate based, at least in part, on the candidate probability; and

displaying the word candidate on a display screen of the computing device.

16. The computer-readable storage medium of claim 15, wherein the computer-executable instructions cause the computing device to:

obtain an input indicating a word; and

reconstruct a freshness value associated with one or more word clusters containing the word, the modification to the freshness value indicating that the one or more word clusters containing the word are more recent than other word clusters of the plurality of word clusters.

17. The computer-readable storage medium of claim 15, wherein the computer-executable instructions cause the computing device to:

receive a text entry;

determine one or more word clusters of the plurality of word clusters associated with the text entry;

obtain a freshness factor associated with the one or more word clusters of the plurality of word clusters associated with the text entry;

obtain a related adjustment factor associated with the one or more word clusters;

obtain a language model value; and

determine the candidate probability associated with the word candidate based, at least in part, on the language model value and the related adjustment factor, wherein the word candidate is selected from individual words associated with the plurality of word clusters.

18. The computer-readable storage medium of claim 17, wherein the computer-executable instructions cause the computing device to:

determine a plurality of word candidates, wherein individual words of the plurality of word candidates comprise an individual candidate probability based, at least in part, on the language model value and the related adjustment factor;

generate data indicating a rank of the word candidate and the individual words of the plurality of word candidates based, at least in part, on the individual candidate probabilities and the candidate probability; and

display data indicating the rank of the word candidate and the individual words.

19. The computer-readable storage medium of claim 15, wherein reconstructing the adjustment factor of the pair of words is also based on a ranking of at least one correlation between the words.

20. The computer-readable storage medium of claim 15, wherein reconstructing the adjustment factor of the pair of words comprises:

determining a cluster density for individual clusters of the plurality of clusters;

determining an ordering of the plurality of clusters based on the cluster density for the individual clusters; and

reconstructing the adjustment factor based on the ordering of the plurality of clusters and the number of common clusters between individual words of the pair of words.