Suggestions based on document topics

• US 9,881,010 B1
• Filed: 05/12/2014
• Issued: 01/30/2018
• Est. Priority Date: 05/12/2014
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Gubin, Maxim, Sung, Sangsoo, Dauber, Kenneth W., Bharat, Krishna
• Primary Examiner(s): Andersen, Kristopher
• Application Number:
  US14/275,403
• Time in Patent Office: 1,359 Days
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 40/166 Editing, e.g. inserting or deleting
• CPC Class Codes: G06F 40/274 Converting codes to words; Guess-ahead of partial word inputs
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• CPC Class Codes: G06F 40/56 Natural language generation

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for providing suggestions based on document topics. In one aspect, a method includes obtaining textual input provided to a document editing application by a user device for inclusion in a document; providing prior text of the document as input to a classification model that provides document topics for the document based on the prior text, and wherein the classification model was trained based on historical documents of other users; and determining, based on the textual input, to provide textual suggestions for inclusion in the document, and in response: selecting candidate suggestions based on the document topics and a suggestion topic for one of the candidate suggestions, each candidate suggestion comprising one or more words; and providing the selected candidate suggestions to the user device for inclusion in the document.

Claims

1. A method implemented by data processing apparatus, the method comprising:

obtaining textual input provided to a document editing application by a user device, the textual input being provided to the document editing application for inclusion in a document;

providing the textual input and prior text of the document as input to a classification model that;

identifies a plurality of document topics for the document based on the prior text and identifies one or more textual input topics for the textual input; and

for each of the plurality of document topics for the document, assigns a document-topic score indicating a probability that the document is related to the document topic to which the document-topic score is assigned, wherein the prior text was included in the document prior to the textual input, and wherein the classification model was trained based on historical documents of other users; and

receiving, from the classification model, one or more textual input topics for the textual input;

determining, based on the textual input, to provide one or more textual suggestions for inclusion in the document, and in response;

searching by document topic in a group of candidate selections to select a plurality of candidate suggestions that are associated with by a confidence score to one or more document topics for the document, the selecting one or more candidate suggestions also based on at least one of the one or more textual input topics and a ranking of each candidate suggestion based on the plurality of document topics and the one or more textual input topics and the one of the one or more topics to which each candidate is associated, and wherein the plurality of candidate suggestions are selected based on the ranking, each candidate suggestion comprising one or more words for inclusion into the document;

generating a topic score for each of the plurality of candidate suggestions that were selected based on i) the selected candidate suggestion'"'"'s confidence score and ii) the document-topic score assigned to the document topic associated with the selected candidate suggestion; and

providing some but not all of the selected candidate suggestions based on the selected candidate suggestions'"'"' topic scores.

2. The method of claim 1, further comprising:

ranking each candidate suggestion based on the topic score for each of the plurality of candidate suggestions that were selected.

3. The method of claim 1, wherein each of the plurality of candidate suggestions that were selected based on at least one of the one or more textual input topics receives a ranking boost relative to candidate suggestions that were not selected based on the at least one of the one or more textual input topics.

4. The method of claim 1, wherein providing the some but not all of the plurality of selected candidate suggestions to the user device for inclusion in the document comprises providing the user device with data that causes one of the selected candidate suggestions to be displayed as an in situ suggestion.

5. The method of claim 1, wherein providing the some but not all of the plurality of selected candidate suggestions to the user device for inclusion in the document comprises providing the user device with data that causes two or more of the selected candidate suggestions to be displayed in a drop-down list within the document.

6. A system comprising:

one or more data processors; and

a data storage device storing instructions that, when executed by the one or more data processors, cause the one or more data processors to perform operations comprising;

obtaining textual input provided to a document editing application by a user device, the textual input being provided to the document editing application for inclusion in a document;

providing the textual input and prior text of the document as input to a classification model that;

identifies a plurality of document topics for the document based on the prior text and identifies one or more textual input topics for the textual input; and

for each of the plurality of document topics for the document, assigns a document-topic score indicating a probability that the document is related to the document topic to which the document-topic score is assigned, wherein the prior text was included in the document prior to the textual input, and wherein the classification model was trained based on historical documents of other users; and

receiving, from the classification model, one or more textual input topics for the textual input;

determining, based on the textual input, to provide one or more textual suggestions for inclusion in the document, and in response;

searching by document topic in a group of candidate selections to select a plurality of candidate suggestions that are associated with by a confidence score to one or more document topics for the document, the selecting the plurality of candidate suggestions also based on at least one of the one or more textual input topics and a ranking of each candidate suggestion based on the plurality of document topics and the one or more textual input topics and the one of the one or more topics to which each candidate is associated, and wherein the plurality of candidate suggestions are selected based on the ranking, each candidate suggestion comprising one or more words for inclusion into the document;

generating a topic score for each of the plurality of candidate suggestions that were selected based on i) the selected candidate suggestion'"'"'s confidence score and ii) the document-topic score assigned to the document topic associated with the selected candidate suggestion; and

providing some but not all of the selected candidate suggestions based on the selected candidate suggestions'"'"' topic scores.

7. The system of claim 6, wherein the operations further comprise:

ranking each candidate suggestion based on the topic score for each of the plurality of candidate suggestions that were selected.

8. The system of claim 6, wherein each of the plurality of candidate suggestions that were selected based on at least one of the one or more textual input topics receives a ranking boost relative to candidate suggestions that were not selected based on the at least one of the one or more textual input topics.

9. The system of claim 6, wherein providing the some but not all of the plurality of selected candidate suggestions to the user device for inclusion in the document comprises providing the user device with data that causes one of the selected candidate suggestions to be displayed as an in situ suggestion.

10. A non-transitory computer readable medium storing instructions that, when executed by one or more data processing apparatus, cause the one or more data processing apparatus to perform operations comprising:

obtaining textual input provided to a document editing application by a user device, the textual input being provided to the document editing application for inclusion in a document;

providing the textual input and prior text of the document as input to a classification model that;

identifies a plurality of document topics for the document based on the prior text and identifies one or more textual input topics for the textual input; and

for each of the plurality of document topics for the document, assigns a document-topic score indicating a probability that the document is related to the document topic to which the document-topic score is assigned, wherein the prior text was included in the document prior to the textual input, and wherein the classification model was trained based on historical documents of other users; and

receiving, from the classification model, one or more textual input topics for the textual input;

determining, based on the textual input, to provide one or more textual suggestions for inclusion in the document, and in response;

searching by document topic in a group of candidate selections to select a plurality of candidate suggestions that are associated with by a confidence score to one or more document topics for the document, the selecting the plurality of candidate suggestions also based on at least one of the one or more textual input topics and a ranking of each candidate suggestion based on the plurality of document topics and the one or more textual input topics and the one of the one or more topics to which each candidate is associated, and wherein the plurality of candidate suggestions are selected based on the ranking, each candidate suggestion comprising one or more words for inclusion into the document;

generating a topic score for each of the plurality of candidate suggestions that were selected based on i) the selected candidate suggestion'"'"'s confidence score and ii) the document-topic score assigned to the document topic associated with the selected candidate suggestion; and

providing some but not all of the selected candidate suggestions based on the selected candidate suggestions'"'"' topic scores.