Extraction of quantitative data from online content

• US 10,242,107 B2
• Filed: 01/11/2015
• Issued: 03/26/2019
• Est. Priority Date: 01/11/2015
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Alonso, Omar, Sellam, Thibault Henri Joseph
• Primary Examiner(s): Wu, Yicun
• Application Number:
  US14/594,154
• Publication Number:
  US 20160203225A1
• Time in Patent Office: 1,535 Days
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

Systems and methods for extracting and organizing quantitative content with a corresponding topic in regard to a particular, desired event, from a body of user posts is presented. More particularly, nearly ubiquitously a body of user posts includes a substantial amount of highly relevant and interesting information goes largely unprocessed and widely inaccessible. According to the disclosed subject matter, a body/corpus of user posts is filtered according to a desired event such that those user posts relating to a desired event is identified. Additionally, the user posts are also filtered according to whether or not the user posts include a quantitative value. An analysis of the filtered user posts is conducted to extract, for qualifying user posts, a quantitative tuple comprising a quantitative value and a corresponding topic.

Claims

1. A quantitative data extraction service as implemented on a computing system, the quantitative data extraction service comprising a processor and a memory, wherein the processor executes instructions retrieved from the memory as part of or in conjunction with additional components to generate quantitative tuples from a corpus of user posts, the additional components comprising:

an executable quantitative tuple generator which, when executing on the processor performs the specialized process of generating quantitative tuples from the corpus of user posts, the quantitative tuple generator coupled to a communications component connected to a network for accessing the corpus of user posts;

an executable event filter that, in execution, filters the corpus of user posts according to a desired event to create a corpus of filtered user posts consisting of a plurality of users posts from the corpus of user posts that relate to the desired event;

an executable quantitative value filter that, in execution, filters each of the corpus of filtered user posts according to whether content of the respective each of the corpus of filtered user posts includes a quantitative value, thereby generating a quantitative event content corpus of a plurality of user posts consisting of user posts from the corpus of filtered user posts having a quantitative value in the respective content and relating to the desired event;

an executable tuple extractor that, in execution, generates from the quantitative event content corpus a body of quantitative tuples corresponding to the desired event, wherein each quantitative tuple of the body of quantitative tuples includes at least a quantitative value extracted from the respective content and a topic, wherein the quantitative value corresponds to the topic; and

wherein the quantitative tuple generator stores the body of quantitative tuples in a quantitative tuple data store.

2. The quantitative data extraction service of claim 1, wherein the event filter component creates the corpus of filtered user posts by determining, for each user post of the corpus of user posts, whether a user post was generated or distributed at a time that corresponds with the desired event.

3. The quantitative data extraction service of claim 2, wherein the event filter component identifies the corpus of filtered user posts by further determining, for each user post of the corpus of user posts, whether the content of a user post includes subject matter that is related to the desired event.

4. The quantitative data extraction service of claim 1, wherein the quantitative value filter component identifies the quantitative event content by determining, for each user post of the corpus of filtered user posts, whether a user post includes numeric characters in the content of the user post.

5. The quantitative data extraction service of claim 4, wherein the quantitative value filter component identifies the quantitative event content by further determining, for each user post of the corpus of filtered user posts, whether a user post includes numeric text in the content of the user post.

6. The quantitative data extraction service of claim 5, wherein the quantitative value filter component identifies quantitative event content by determining, for each user post of the corpus of filtered user posts, whether a user post includes misspelled numeric text in the content of the user post.

7. The quantitative data extraction service of claim 4, wherein the quantitative value filter component identifies the quantitative event content by determining, for each user post of the second corpus of filtered user posts, whether a user post includes a non-numeric usage of numeric characters in the content of the user post.

8. The quantitative data extraction service of claim 1 further comprising a tuple clustering component aggregating from the body of quantitative tuples similar quantitative tuples into clusters of like quantitative tuples.

9. The quantitative data extraction service of claim 8 further comprising a topical organizing component generating a topical arrangement for the desired event according to the topics of the clustered quantitative tuples in a topical organizing component of the quantitative tuple generator.

10. The quantitative data extraction service of claim 1, wherein the tuple extraction component extracts additional elements from each of the user posts in the quantitative event content corpus related to the quantitative value and corresponding topic, the additional elements comprising at least one of:

contextual information regarding the quantitative tuple not including the quantitative value or the corresponding topic;

a confidence value regarding validity of information in the quantitative tuple with regard to the desired event;

a date of the user post;

that the user post was generated by an authority of the desired event;

a location in which the user post was generated; and

how many user posts are directed to the corresponding topic of the quantitative tuple.

11. A computer implemented method which, when operated on a computer processor, creates a specialized system for generating quantitative tuples, the method comprising:

accessing, through a network communication component of a computer system having the computer processor, multiple user posts found on the internet by a quantitative tuple generator, the quantitative tuple generator creating a corpus of user posts from the found user posts;

filtering the corpus of user posts according to a desired event through an event filter component to create a first subset of the corpus of user posts, the first subset comprising a plurality of users posts from the corpus of user posts that have content in the post placed there by the user that relates to the desired event;

responsive to filtering the corpus of user posts, filtering each post of the first subset through a quantitative value filter component according to whether user placed content of the respective each post of the first subset includes a quantitative value, thereby generating a second subset of user posts comprising user posts from the first subset having a quantitative value in the respective content and relating to the desired event;

responsive to filtering each post of the first subset, extracting from the second subset a body of quantitative tuples corresponding to the desired event through a tuple extraction component, wherein each of the quantitative tuples of the body of quantitative tuples includes at least a quantitative value extracted from the respective content and a corresponding topic; and

storing the body of quantitative tuples in a quantitative tuple data store.

12. The method of claim 11, wherein filtering the first corpus of user posts according to the desired event comprises, for each user post of the first corpus of user posts, determining whether a user post was generated or distributed at a time that corresponds with the desired event.

13. The method of claim 12, wherein filtering the corpus of user posts comprises, for each user post of the corpus of user posts, determining whether the content of a user post includes subject matter that is related to the desired event.

14. The method of claim 11, wherein filtering each post of the first subset comprises, for each user post of the first subset, determining any one or more of:

a user post includes numeric characters in the content of the user post;

a user post includes numeric text in the content of the user post;

a user post includes misspelled numeric text in the content of the user post; and

a user post includes a non-numeric usage of numeric characters in the content of the user post.

15. The method of claim 11 further comprising aggregating similar quantitative tuples from the body of quantitative tuples into clusters of like quantitative tuples.

16. The method of claim 15 further comprising generating a topical arrangement for the desired event according to topics and subtopics of the clustered quantitative tuples.

17. The method of claim 11 further comprising extracting additional elements related to the quantitative value and corresponding topic of at least one quantitative tuple, the additional elements comprising at least one of:

contextual information regarding the quantitative tuple not including the quantitative value or the corresponding topic;

a confidence value regarding validity of information in the quantitative tuple with regard to the desired event;

a date of the user post;

that the user post was generated by an authority of the desired event;

a location in which the user post was generated; and

how many user posts are directed to the corresponding topic of the quantitative tuple.

18. The method of claim 11, wherein the content of the corpus of user posts comprises any one of textual content, audio content, image content, and video content.

19. A computer readable medium bearing computer executable instructions which, when executed on a computing system comprising at least a processor and a memory, carry out a specialized method for generating quantitative tuples, the method comprising:

accessing, through a network communication component of the computer system and executed on the processor of the computer system, a plurality of user posts retrieved from the internet by a quantitative tuple generator, the quantitative tuple generator creating a corpus of user posts from a subset of the retrieved user posts;

filtering the corpus of user posts through an event filter component to create a first subset of user posts that were generated in response to occurrence of an event, the first subset comprising a plurality of users posts from the corpus of user posts each having content placed in the post by a user that relates to the event;

filtering each post in the first subset through a quantitative value filter component to create a second subset of user posts, each of which includes a quantitative value in the respective user content relating to the event;

extracting from content of the posts in the second subset a body of quantitative tuples corresponding to the event through a tuple extraction component, wherein each quantitative tuple of the body of quantitative tuples of the body of quantitative tuples includes at least a quantitative value and a corresponding topic; and

storing the body of quantitative tuples in a quantitative tuple data store.

20. The computer readable medium of claim 19, wherein the method further comprises:

aggregating similar quantitative tuples from the body of quantitative tuples into clusters of like quantitative tuples; and

generating a topical arrangement for the desired event according to the topics of the clustered quantitative tuples.