Systems and methods for automated content generation

• US 10,025,770 B2
• Filed: 06/03/2013
• Issued: 07/17/2018
• Est. Priority Date: 05/13/2010
• Status: Active Grant

• Current Assignee: Expedia Inc. (Expedia Group, Inc.)
• Original Assignee: Expedia Inc. (Expedia Group, Inc.)
• Inventors: Waksberg, Rene, Hsu, Donny, Bradley, Patrick
• Primary Examiner(s): GUERRA-ERAZO, EDGAR X
• Application Number:
  US13/908,767
• Publication Number:
  US 20130346856A1
• Time in Patent Office: 1,870 Days
• Field of Search: 704 1- 10, 707709, 707748-751, 707828-831, 707E17001-E17143
• CPC Class Codes: G06F 16/972 Access to data in other repository systems, e.g. legacy data or dynamic Web page generation
• CPC Class Codes: G06F 40/186 Templates

Abstract

A content generation service is described that generates content for electronic documents in different languages based upon templates. The templates may include paragraph templates composed of sentence types including sentence templates. The sentence templates may further include variables having corresponding attributes. Each of the paragraph templates, sentence templates, and attributes may be hierarchically organized. The content generation service may obtain data describing an item of interest, such as a travel item. The obtained data may further specify a document language, section and paragraph for which content is to be generated. Content is generated for variables in hierarchical order, with higher ranked paragraphs considered first. Within the highest ranked paragraph, a sentence type is selected and the variables within the highest ranked sentence template of the sentence type are considered. Sentence templates which contain attributes matched by the data obtained for the item of interest are added to the generated document.

Claims

1. A system for generating content regarding one or more content subjects utilizing structured information in a database, the system comprising:

a data store configured to store a template hierarchy comprising;

a plurality of paragraph templates, wherein each of the plurality of paragraph templates comprises one or more sentence types and is associated with a natural language of one or more natural languages;

a plurality of sentence templates, wherein each of the plurality of sentence templates corresponds to at least one of the one or more sentence types of at least one of the plurality of paragraph templates, and wherein each of the plurality of sentence templates comprises one or more variables;

wherein the plurality of paragraph templates and the plurality of sentence templates are ranked within the template hierarchy independently of attributes of individual content subjects and at least according to a desirability of use; and

one or more processors in communication with the data store, wherein the one or more processors are configured to;

obtain structured information from the database comprising a set of attributes of a content subject, wherein the set of attributes is expressed in a first natural language of the one or more natural languages;

identify, based at least in part on the set of attributes and a ranking of the plurality of paragraph templates within the template hierarchy, a first paragraph template from the plurality of paragraph templates;

for each sentence type of the first paragraph template, identify, based at least in part on the set of attributes and on a ranking of the plurality of sentence templates within the template hierarchy, a sentence template from the plurality of sentence templates and corresponding to the sentence type, which sentence template includes a variable that matches an attribute within the set of attributes; and

generate content in the first natural language regarding the first content subject based at least in part on processing;

(i) the first paragraph template and the identified sentence templates from the template hierarchy which each include a variable that matches an attribute within the set of attributed included within the structured information with (ii) at least a portion of the one or more attributes within the structured information from the database.

2. The system of claim 1, wherein the content subject is a property.

3. The system of claim 2, wherein the set of attributes comprises at least one property name, property amenity, property rating, property location, or point of interest.

4. The system of claim 1, wherein the one or more natural languages comprise a plurality of natural languages.

5. The system of claim 4, wherein the one or more processors are configured to identify the first paragraph template based at least in part on the first natural language.

6. The system of claim 1, wherein the first paragraph template further comprises an optional sentence type.

7. The system of claim 6, wherein the one or more processors are further configured to:

identify a sentence template from the plurality of sentence templates corresponding to the optional sentence type, wherein the sentence template comprises one or more variables; and

determine that each of the one or more variables of the sentence template corresponding to the optional sentence type corresponds to at least one attribute of the set of attributes;

wherein the content is generated based at least in part on the sentence template corresponding to the optional sentence type.

17. The system of claim 1. wherein the variable matches the attribute within the set of attributes when the attribute satisfies the variable.

8. A computer-implemented method for generating content regarding one or more content subjects utilizing structured information in a database, the computer-implemented method comprising:

under control of one or more computing devices configured with specific computer-executable instructions,

obtaining structured information from the database comprising a set of attributes of a content subject, wherein the set of attributes is expressed in a first natural language;

accessing information identifying a template hierarchy, the template hierarchy comprising;

a plurality of paragraph templates, wherein each of the plurality of paragraph templates comprises one or more sentence types and is associated with a natural language of one or more natural languages;

a plurality of sentence templates, wherein each of the plurality of sentence templates corresponds to at least one of the one or more sentence types of at least one of the plurality of paragraph templates, and wherein each of the plurality of sentence templates comprises one or more variables;

wherein the plurality of paragraph templates and the plurality of sentence templates are ranked within the template hierarchy independently of attributes of individual content subjects and at least according to a desirability of use;

identifying a first paragraph template from the plurality of paragraph templates based at least in part on the set of attributes and a ranking of the plurality of paragraph templates within the template hierarchy, and wherein identifying the first paragraph template comprises, for each sentence type of the first paragraph template;

identifying a sentence template from the plurality of sentence templates, wherein the sentence template corresponds to the sentence type, and wherein the sentence template comprises one or more variables; and

determining that each of the one or more variables matches at least one attribute of the set of attributes; and

generating content in the first natural language based at least in part on processing;

(ii) the first paragraph template and the identified sentence templates from the template hierarchy which each include a variable that matches an attribute within the set of attributes included within the structured information with (ii) at least a portion of the one or more attributes within the structured information from the database.

9. The computer-implemented method of claim 8, wherein the set of attributes comprise at least one attribute type and at least one corresponding attribute value.

10. The computer-implemented method of claim 9, wherein each sentence template of the one or more sentence templates includes an indicator specifying whether variables within the sentence template match an attribute type or attribute value.

11. The computer-implemented method of claim 8, wherein the generated electronic document does not include an unsatisfied sentence template, and wherein the unsatisfied sentence template has at least one variable that does not match an attribute within the set of attributes.

12. The computer-implemented method of claim 8, wherein the first paragraph template further comprises an optional sentence type.

13. A non-transitory, computer-readable medium including computer-executable instructions for generating content regarding one or more content subjects utilizing structured information in a database, the computer-executable instructions comprising:

a first set of computer-executable instructions that, when executed by one or more computing devices, cause the one or more computing devices to maintain a template hierarchy comprising a plurality of paragraph templates and a plurality of sentence templates, wherein;

each of the plurality of paragraph templates is associated with a natural language of one or more natural languages,

each of the plurality of paragraph templates comprises one or more sentence types,

each of the plurality of sentence templates corresponds to at least one of the one or more sentence types in at least one of the plurality of paragraph templates,

each of the plurality of sentence templates comprises one or more variables; and

the plurality of paragraph templates and the plurality of sentence templates are ranked within the template hierarchy independently of attributes of individual content subjects and at least according to a desirability of use; and

a second set of computer-executable instructions that, when executed by the one or more computing devices, cause the one or more computing devices to;

obtain structured information from the database comprising a set of attributes of a content subject, wherein the set of attributes is expressed in a first natural language of the one or more natural languages;

identify, based at least in part the set of attributes and a ranking of the plurality of paragraph templates within the template hierarchy, a first paragraph template from the plurality of paragraph templates;

for each sentence type of the first paragraph template, identify, based at least in part on the set of attributes and on a ranking of the plurality of sentence templates within the template hierarchy, a sentence template from the plurality of sentence templates that corresponds to the sentence type, which sentence template includes a variable that matches an attribute within the set of attributes; and

generate content in the first natural language based at least in part on processing;

(i) the first paragraph template and the identified sentence templates from the template hierarchy which each include a variable that matches an attribute within the set of attributes included within the structured information with (ii) with at least a portion of the set of attributes within the structured information from the database.

14. The non-transitory, computer-readable medium of claim 13, wherein the one or more natural languages comprise a plurality of natural languages.

15. The non-transitory, computer-readable medium of claim 14, wherein the first paragraph template is further identified based at least in part on the natural language.

16. The non-transitory, computer-readable medium of claim 13, wherein the first paragraph template further comprises an optional sentence type.