Method and apparatus for identifiying similar questions in a consultation system

• US 9,646,079 B2
• Filed: 05/04/2012
• Issued: 05/09/2017
• Est. Priority Date: 05/04/2012
• Status: Active Grant

• Current Assignee: Pearl.com LLC (JustAnswer LLC)
• Original Assignee: Pearl.com LLC (JustAnswer LLC)
• Inventors: Bierner, Gann, Cooper, Edwin
• Primary Examiner(s): Hutton, Nan
• Application Number:
  US13/464,230
• Publication Number:
  US 20130297625A1
• Time in Patent Office: 1,831 Days
• Field of Search: 707754
• CPC Class Codes: G06F 16/33 Querying
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Embodiments of the present invention further provide systems and methods for automatically identifying questions on topics similar to a newly submitted question to an online the consultation system.

Claims

1. A method comprising:

receiving at an online consultation system a question associated with a predetermined category;

extracting a plurality of candidate topics from the received question;

applying a first term frequency inverse document frequency (TFIDF) filter to the extracted plurality of candidate topics to identify a first sorted list of the plurality of extracted candidate topics according to an affinity of the candidate topics to the predetermined category;

applying a second TFIDF filter to the first sorted list to identify a second sorted list of the plurality of extracted candidate topics according to an affinity of the candidate topics to the received question;

identifying as similar questions a plurality of previously submitted questions having topics matching the second sorted list of the plurality of extracted candidate topics;

creating excerpts of the identified similar questions by selecting one or more sections of each of the identified similar questions;

presenting to one or more users via the online consultation system the excerpts of the identified plurality of previously submitted questions

while the user is waiting for an answer to the question.

2. The method of claim 1, wherein the similar questions are made available to subscription users.

3. The method of claim 1, wherein after reviewing similar questions, the one or more users have the opportunity of accepting an answer to one of the plurality of previously submitted questions for a reduced fee.

4. The method of claim 1, wherein after reviewing similar questions, the one or more users are given an option to modify the received question.

5. The method of claim 1, wherein subscription users are allowed to search a database of previously submitted questions to identify questions including similar topics.

6. The method of claim 1, wherein the first sorted list of the plurality of extracted candidate topics ranks candidate topics occurring more often inside the predetermined category that outside the predetermined category higher than candidate topics occurring less often inside the predetermined category that outside the predetermined category.

7. The method of claim 1, wherein the second sorted list of the plurality of extracted candidate topics re-ranks the first ranking, where the second sorted list of the plurality of extracted candidate topics ranks candidate topics occurring more often inside the received question that within other questions in the predetermined category higher than candidate topics occurring less often inside the received question that within other questions in the predetermined category.

8. The method of claim 1, wherein extracting the plurality of candidate topics from the received question includes automatically performing, by a linguistic engine, sentence detection, tokenization, phrase extraction, and tagging of speech parts for the received question.

9. A topic extraction apparatus in an online consultation system, the apparatus comprising a plurality of engines for:

receiving at an online consultation system a question associated with a predetermined category;

extracting a plurality of candidate topics from the received question;

applying a first term frequency inverse document frequency (TFIDF) filter to the extracted plurality of candidate topics to identify a first sorted list of the plurality of extracted candidate topics according to an affinity to the predetermined category;

applying a second TFIDF filter to the first sorted list to identify a second sorted list of the plurality of extracted candidate topics according to an affinity to the received question;

identifying as similar questions a plurality of previously submitted questions having topics matching the second sorted list of the plurality of extracted candidate topics;

creating excerpts of the identified similar questions by selecting one or more sections of each of the identified similar questions;

presenting to one or more users via the online consultation system the excerpts of the identified plurality of previously submitted questions while the user is waiting for an answer to the question.

10. A non-transitory machine-readable storage medium having embodied thereon instructions which when executed by at least one processor, causes a machine to perform operations comprising:

receiving at an online consultation system a question associated with a predetermined category;

extracting a plurality of candidate topics from the received question;

applying a first term frequency inverse document frequency (TFIDF) filter to the extracted plurality of candidate topics to identify a first sorted list of the plurality of extracted candidate topics according to an affinity to the predetermined category;

applying a second TFIDF filter to the first sorted list to identify a second sorted list of the plurality of extracted candidate topics according to an affinity to the received question;

identifying as similar questions a plurality of previously submitted questions having topics matching the second sorted list of the plurality of extracted candidate topics;

creating excerpts of the identified similar questions by selecting one or more sections of each of the identified similar questions;

presenting to one or more users via the online consultation system the excerpts of the identified plurality of previously submitted questions while the user is waiting for an answer to the question.