Enhanced document input parsing

• US 10,430,469 B2
• Filed: 10/04/2016
• Issued: 10/01/2019
• Est. Priority Date: 06/27/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bastide, Paul R., Broomhall, Matthew E., Loredo, Robert E., Lu, Fang
• Primary Examiner(s): Pham, Michael
• Application Number:
  US15/284,928
• Publication Number:
  US 20170024374A1
• Time in Patent Office: 1,092 Days
• Field of Search: 707755, 707748
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• 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 40/205 Parsing
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

An approach is provided for an information handling system that includes a processor and a memory to analyze documents. In the approach, an electronic document is received with the document including content, such as text, and revision metadata that is associated with the content. The revision metadata is analyzed and the approach identifies a confidence level based on the analysis. The confidence level is associated with the electronic document content. The confidence level can then be utilized by a Question and Answer (QA) system.

Claims

1. A method, in an information handling system comprising a processor and a memory, of analyzing documents, the method comprising:

receiving, at a question answer system, an electronic document that includes a content and a revision metadata generated by at least one identified revision author;

retrieving, from an endorser at the question answer system, endorsement data that endorses the identified revision author, wherein the endorser is different than the identified revision author;

identifying, by the question answer system, a topic area associated with the electronic document;

collecting, from one or more network sources by the question answer system, expertise-related data corresponding to the revision author relative to the identified topic area;

determining, by the question answer system, a revision author expertise level of the identified revision author based upon the endorsement data and the expertise-related data;

identifying, by the question answer system, a confidence level based on the revision author expertise level;

assigning, by the question answer system, the confidence level to the electronic document content; and

in response to determining that the confidence level of the electronic document is higher than a different confidence level of a different document;

selecting, by the question answer system, the electronic document over the different electronic document; and

generating, by the question answer system, one or more answers to a question based, at least in part, on the selected electronic document.

2. The method of claim 1 further comprising:

parsing the electronic document into a plurality of sections; and

analyzing each of the plurality of sections based on a corresponding one of a plurality of revisions included in the revision metadata.

3. The method of claim 1 wherein the identified revision author corresponds to a selected one of a plurality of revisions in the revision metadata.

4. The method of claim 3 further comprising:

identifying a revision type associated with the selected revision, wherein the confidence level is based on the revision type.

5. The method of claim 1 further comprising:

determining an endorser expertise of the endorser based on the endorsement data and the identified topic area, wherein the revision author expertise level is adjusted based on the endorser expertise.

6. The method of claim 1 further comprising:

identifying at least one document author associated with the electronic document;

collecting author expertise-related data pertaining to the identified document author from at least one of the one or more network sources; and

determining a document author expertise level associated with the identified document author based on the collected author expertise-related data, wherein the confidence level is adjusted based on the document author expertise level.