INFORMATION IDENTIFICATION AND EXTRACTION

• US 20170235726A1
• Filed: 02/12/2016
• Published: 08/17/2017
• Est. Priority Date: 02/12/2016
• Status: Abandoned Application

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: UCHINO, Kanji, WANG, Jun
• Application Number:
  US15/043,406
• Publication Number:
  US 20170235726A1
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/38 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

A computer implemented method of information identification and extraction may include creating an author object in a database for each author of multiple digital documents. For each author object created, the computer implemented method may also include obtaining an indication of social media accounts in a social media based on a search in the social media for a name of the author in the author object. Alternately or additionally, for each social media account obtained through the search of the social media, the method may include determining whether the social media account is associated with the author of the author object based on two or more of the following: a name score, a profile score, a content score, and an interaction score.

Claims

1. A computer implemented method of information identification and extraction, the method comprising:

creating an author object in a database for each author of a plurality of digital documents;

for each author object created, the computer implemented method includes;

obtaining an indication of social media accounts in a social media based on a search in the social media for a name of the author in the author object; and

for each social media account obtained through the search of the social media, the computer implemented method includes;

generating a name score based on a comparison of a name from the author object and a social media name from a social media account object generated based on the social media account;

generating a profile score based on a comparison of author profile data from the author object and social media profile data from the social media account object;

generating a content score based on a comparison of topics from postings on the social media account and topics for each of the digital documents associated with the author from the author object;

generating an interaction score based on an evaluation of social connections in the social media account and co-authors for each of the digital documents associated with the author from the author object; and

determining if the social media account is associated with the author of the author object based on the name score, the profile score, the content score, and the interaction score;

extracting data from new posts from the social media accounts associated with the authors of each of the author objects; and

providing the data in an organization based on the topics of the digital documents.

2. The computer implemented method of claim 1, wherein the author profile data includes one or more of a title of the author, an affiliation of the author, an expertise of the author, and a location of the author.

3. The computer implemented method of claim 1, wherein comparison of the author profile data and the social media profile data includes:

constructing an author vector using the author profile data;

constructing a social media vector using the social media profile data; and

calculating a similarity between the author vector and the social media vector, wherein the calculated similarity is the profile score.

4. The computer implemented method of claim 1, further comprising determining the topics from the postings on the social media account, wherein determining the topics includes:

removing the postings shorter than a threshold number of words;

obtaining content from embedded links in the postings;

aggregating the content; and

determining topic distribution of the aggregating content.

5. The computer implemented method of claim 1, wherein determining if the social media account is associated with the author of the author object based on the name score, the profile score, the content score, and the interaction score includes:

assigning each of the name score, the profile score, the content score, and the interaction score a weight;

linearly combining the weighted name score, the weighted profile score, the weighted content score, and the weighted interaction score; and

applying the linear combination to a machine learning algorithm to determine if the social media account is associated with the author of the author object.

6. The computer implemented method of claim 1, further comprising:

obtaining the plurality of digital documents from one or more web sites; and

determining a topic of each of the digital documents using a topic model analysis.

7. The computer implemented method of claim 1, wherein creating the author object includes extracting the name, the author profile data, and the co-authors from the digital documents.

8. A non-transitory computer-readable storage media including computer-executable instructions configured to cause a system to perform operations, the operations comprising:

create an author object in a database for each author of a plurality of digital documents;

for each author object created, the operations include;

obtain an indication of social media accounts in a social media based on a search in the social media for a name of the author in the author object; and

for each social media account obtained through the search of the social media, determine whether the social media account is associated with the author of the author object based on two or more of the following;

a name score, a profile score, a content score, and an interaction score, wherein;

the name score is generated based on a comparison of a name from the author object and a social media name from a social media account object generated based on the social media account,

the profile score is generated based on a comparison of author profile data from the author object and social media profile data from the social media account object,

the content score is generated based on a comparison of topics from postings on the social media account and topics for each of the digital documents associated with the author from the author object, and

the interaction score is generated based on an evaluation of social connections in the social media account and co-authors for each of the digital documents associated with the author from the author object.

9. The non-transitory computer-readable storage media of claim 8, wherein the author profile data includes one or more of a title of the author, an affiliation of the author, an expertise of the author, and a location of the author.

10. The non-transitory computer-readable storage media of claim 8, wherein comparison of the author profile data and the social media profile data includes:

construct an author vector using the author profile data;

construct a social media vector using the social media profile data; and

calculate a similarity between the author vector and the social media vector, wherein the calculated similarity is the profile score.

11. The non-transitory computer-readable storage media of claim 8, wherein the operations further comprise determine the topics from the postings on the social media account, wherein determine the topics includes:

remove the postings shorter than a threshold number of words;

obtain content from embedded links in the postings;

aggregate the content; and

determine topic distribution of the aggregated content.

12. The non-transitory computer-readable storage media of claim 8, wherein creation of the author object includes extract the name, the author profile data, and the co-authors from the digital documents.

13. The non-transitory computer-readable storage media of claim 8, wherein determine if the social media account is associated with the author of the author object based on the name score, the profile score, the content score, and the interaction score includes:

assign each of the name score, the profile score, the content score, and the interaction score a weight;

linearly combine the weighted name score, the weighted profile score, the weighted content score, and the weighted interaction score; and

apply the linear combination to a machine learning algorithm to determine if the social media account is associated with the author of the author object.

14. The non-transitory computer-readable storage media of claim 8, wherein create the author object includes extracting the name, the author profile data, and the co-authors from the digital documents.

15. A computer implemented method of information identification and extraction, the method comprising:

creating an author object in a database for each author of a plurality of digital documents;

for each author object created, the computer implemented method includes;

obtaining an indication of social media accounts in a social media based on a search in the social media for a name of the author in the author object; and

for each social media account obtained through the search of the social media, determining whether the social media account is associated with the author of the author object based on two or more of the following;

a name score, a profile score, a content score, and an interaction score, wherein;

the name score is generated based on a comparison of a name from the author object and a social media name from a social media account object generated based on the social media account,

the profile score is generated based on a comparison of author profile data from the author object and social media profile data from the social media account object,

the content score is generated based on a comparison of topics from postings on the social media account and topics for each of the digital documents associated with the author from the author object, and

the interaction score is generated based on an evaluation of social connections in the social media account and co-authors for each of the digital documents associated with the author from the author object.

16. The computer implemented method of claim 15, wherein the author profile data includes one or more of a title of the author, an affiliation of the author, an expertise of the author, and a location of the author.

17. The computer implemented method of claim 15, wherein comparison of the author profile data and the social media profile data includes:

constructing an author vector using the author profile data;

constructing a social media vector using the social media profile data; and

calculating a similarity between the author vector and the social media vector, wherein the calculated similarity is the profile score.

18. The computer implemented method of claim 15, further comprising determining the topics from the postings on the social media account, wherein determining the topics includes:

removing the postings shorter than a threshold number of words;

obtaining content from embedded links in the postings;

aggregating the content; and

determining topic distribution of the aggregated content.

19. The computer implemented method of claim 15, wherein determining if the social media account is associated with the author of the author object based on the name score, the profile score, the content score, and the interaction score includes:

assigning each of the name score, the profile score, the content score, and the interaction score a weight;

linearly combining the weighted name score, the weighted profile score, the weighted content score, and the weighted interaction score; and

applying the linear combination to a machine learning algorithm to determine if the social media account is associated with the author of the author object.

20. The computer implemented method of claim 15, wherein creating the author object includes extracting the name, the author profile data, and the co-authors from the digital documents.