NLP duration and duration range comparison methodology using similarity weighting

• US 9,535,980 B2
• Filed: 03/07/2016
• Issued: 01/03/2017
• Est. Priority Date: 02/19/2014
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Contreras, David, Sizemore, Robert C., Smith, Sterling R.
• Primary Examiner(s): Hill, Stanley K
• Assistant Examiner(s): Chubb, Mikayla
• Application Number:
  US15/062,497
• Publication Number:
  US 20160189033A1
• Time in Patent Office: 302 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 16/243 Natural language query formulation
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition

Abstract

A mechanism is provided in a question answering system for duration similarity weighting in a question answering system. The question answering system receives an input question. The question answering system generates a plurality of candidate answers to the input question using a corpus of information. The question answering system identifies a question-side duration value for the input question and identifies a policy-side duration value for a policy to be applied to an entity identified in the input question. The question answering system determines a duration similarity weight for each of the plurality of candidate answers. The question answering system applies the duration similarity weight to each of the plurality of candidate answers. The question answering system ranks the plurality of candidate answers and outputs one or more of the candidate answers.

Claims

1. A method, in a data processing system, for duration similarity weighting in a question answering system, the method comprising:

receiving an input question concerning an entity;

generating a plurality of candidate answers to the input question using a corpus of information;

identifying a question-side duration value for the entity identified in the input question;

identifying a policy-side duration value for a policy applied to the entity to generate each of the plurality of candidate answers;

determining a duration similarity weight for each of the plurality of candidate answers based on a similarity between the question-side duration value and the policy-side duration value;

applying the duration similarity weight to each of the plurality of candidate answers;

ranking the plurality of candidate answers based on applying the duration similarity weight to each of the plurality of candidate answers; and

outputting one or more of the candidate answers based on the ranking.

2. The method of claim 1, wherein applying the duration similarity weight to each of the plurality of candidate answers comprises:

determining a confidence score for each given candidate answer; and

multiplying the duration similarity weight of the given candidate answer by the confidence score of the given candidate answer to form a duration similarity weighted confidence score.

3. The method of claim 2, wherein ranking the plurality of candidate answers comprises ranking the plurality of candidate answers according to duration similarity weighted confidence score.

4. The method of claim 1, wherein identifying the question-side duration value for the input question comprises identifying the question-side duration value within text of the input question.

5. The method of claim 1, wherein identifying the question-side duration value for the input question comprises:

identifying question-side support evidence from the corpus of information for the input question; and

identifying the question-side duration value within the question-side support evidence.

6. The method of claim 1, wherein identifying the policy-side duration value comprises:

identifying policy-side support evidence for the policy to be applied to the entity identified in the input question; and

identifying the policy-side duration value within the policy-side support evidence.

7. The method of claim 1, wherein determining the duration similarity weight for a given candidate answer comprises calculating the duration similarity weight, similarity, using the following equation:

8. The method of claim 1, wherein determining the duration similarity weight for a given candidate answer comprises calculating the duration similarity weight, similarity, using the following equation: