Method for dynamically assigning question priority based on question extraction and domain dictionary

• US 10,049,153 B2
• Filed: 03/06/2015
• Issued: 08/14/2018
• Est. Priority Date: 09/30/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chandrasekaran, Swaminathan, Haverlock, Kevin B., Hill, Nicholas J., Mobley, Jr., Roy D.
• Primary Examiner(s): Kuddus, Daniel
• Application Number:
  US14/640,373
• Publication Number:
  US 20160092792A1
• Time in Patent Office: 1,257 Days
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/36 Creation of semantic tools, e.g. ontology or thesauri
• CPC Class Codes: G06F 17/3053 using ranking
• CPC Class Codes: G06F 17/30654 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 17/30684 using natural language analysis
• CPC Class Codes: G06F 17/30731 Creation of semantic tools
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/046 Forward inferencing; Production systems
• CPC Class Codes: G06N 99/005 Learning machines, i.e. computer in which a programme is changed according to experience gained by the machine itself during a complete run neural networks G06N3/02; knowledge based models G06N5/00; fuzzy logic systems G06N7/02; adaptive control systems G05B13/00
• CPC Class Codes: H04L 67/306 User profiles

Abstract

An approach is provided dynamically prioritizing question requests based on extracted question data. In the approach, performed by an information handling system, a number of question requests to a question and answering (QA) system are received from a computer network, and a plurality of question priority parameters are identified, including one or more question topics and a plurality question context parameters, by performing natural language processing (NLP) analysis of each question request. The approach determines a target priority value for each question request based on the plurality of question priority parameters identified for said question request. By evaluating the target priority values for the plurality of question requests, processing of the question requests is prioritized by applying an artificial intelligence (AI) learned models and rule-based logic at the information handling system to evaluate the target priority values for the plurality of question requests.

Claims

1. A method, in an information handling system comprising a processor and a memory, of dynamically prioritizing questions for response processing based on extracted question parameters, the method comprising:

receiving, by the information handling system, a plurality of questions;

extracting, by the information handling system, a plurality of question priority parameters comprising one or more question topics and a plurality of question context parameters for each question;

identifying, by the information handling system, system performance parameters for the information handling system comprising processor utilization data, available memory space data, and bandwidth utilization data;

determining, by the information handling system, a target priority value for each question based on the system performance parameters and the plurality of question priority parameters identified for said question;

dynamically prioritizing, by the information handling system, response processing of the plurality of questions by comparing the target priority values for each of the plurality of questions to generate a prioritized plurality of questions;

storing, by a prioritization manager subsystem of the information handling system, the prioritized plurality of questions in a question queue at the information handling system which is configured to support asynchronous delivery of the prioritized plurality of questions; and

passing the prioritized plurality of questions as messages between one or more question answering systems such that the prioritization manager subsystem and the one or more question answering systems do not need to interact with the question queue at the same time.

2. The method of claim 1, wherein extracting the plurality of question priority parameters comprises performing, by the information handling system, a natural language processing (NLP) analysis of each question, wherein the NLP analysis results in one or more question topics comprising extracted named entity information from said question.

3. The method of claim 2, wherein extracting the plurality of question priority parameters further comprises comparing the extracted named entity information from each question to key priority terms contained in one or more domain dictionary databases to identify extracted named entity information that matches one or more key priority terms.

4. The method of claim 1, wherein extracting the plurality of question priority parameters comprises identifying, by the information handling system, for each question, question timing data, question source data, question location data, profile data for the user or user group submitting the question and performance requirement data for the user or user group submitting the question.

5. The method of claim 1, wherein determining a target priority value for each question comprises executing, by the information handling system, a priority prediction machine learning model against the plurality of question priority parameters identified for each question to generate a priority value for said question.

6. The method of claim 1, further comprising quantifying, by the information handling system, one or more priority attribute features for each question based on the system performance parameters to indicate operational performance-related data for the information handling system.

7. The method of claim 1, wherein dynamically prioritizing response processing of the plurality of questions comprises applying an artificial intelligence (AI) learned models and rule-based logic at the information handling system to sort the plurality of questions based on the target priority values for the plurality of questions and to present the plurality of questions in sorted order to a question answering (QA) system for response processing.