Caching natural language questions and results in a question and answer system

• US 9,318,027 B2
• Filed: 09/19/2014
• Issued: 04/19/2016
• Est. Priority Date: 06/20/2013
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Byron, Donna K., Krishnamurthy, Lakshminarayanan, Seabolt, Edward E.
• Primary Examiner(s): Hill, Stanley K
• Assistant Examiner(s): Smith, Paulinho E
• Application Number:
  US14/491,344
• Publication Number:
  US 20150044660A1
• Time in Patent Office: 578 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 16/338 Presentation of query results
• CPC Class Codes: G09B 7/00 Electrically-operated teaching apparatus or devices working with questions and answers mechanically operated G09B3/00; computing arrangements G06F
• CPC Class Codes: G09B 7/02 of the type wherein the student is expected to construct an answer to the question which is presented or wherein the machine gives an answer to the question presented by a student

Abstract

Mechanisms are provided for answering an input question. An input question to be answered from a source is received and processed to one or more extract features of the input question. The extracted one or more features are compared to cached features stored in one or more entries of a question and answer (QA) cache. A determination is made as to whether there is a matching entry in the one or more entries of the QA cache based on results of the comparing and, if so, candidate answer information is retrieved from the matching entry. The retrieved candidate answer information is returned to the source of the input question as candidate answer information for answering the input question.

Claims

1. A method, in a data processing system comprising a processor and a memory, for answering an input question, the method comprising:

receiving, in the data processing system, an input question to be answered from a source;

processing, by the data processing system, the input question to extract one or more features of the input question;

comparing, by the data processing system, the extracted one or more features to cached features stored in one or more entries of a question and answer (QA) cache of the data processing system;

determining, by the data processing system, whether there is a matching entry in the one or more entries of the QA cache based on results of the comparing, wherein determining whether there is a matching entry in the one or more entries of the QA cache comprises;

generating, for each entry in the QA cache, a match value indicative of a degree of matching between the one or more extracted features of the input question to cached features of the entry in the QA cache; and

comparing the match value to one or more threshold values indicating one or more requisite degrees of similarity between the input question and an entry in the QA cache, wherein;

in response to the match value equaling or exceeding a first threshold value, a corresponding entry is determined to match the input question, and

in response to the match value being less than the first threshold value but the match value being equal to or greater than a second threshold value, determining that the corresponding entry is sufficiently similar for updating the corresponding entry with the one or more extracted features of the input question;

retrieving, by the data processing system, in response to a matching entry being present in the one or more entries of the QA cache, candidate answer information from the matching entry; and

returning, by the data processing system, the retrieved candidate answer information to the source of the input question as candidate answer information for answering the input question.

2. The method of claim 1, wherein, in response to determining that there is a matching entry in the one or more entries of the QA cache, processing of the input question by a predetermined set of QA system pipeline stages is circumvented.

3. The method of claim 1, further comprising:

in response to determining that there is not a matching entry in the one or more entries of the QA cache, processing the input question through a QA system pipeline to generate one or more generated candidate answers for answering the input question.

4. The method of claim 3, further comprising:

in response to determining that there is not a matching entry in the one or more entries of the QA cache, generating a new cache entry in the QA cache for the input question, wherein the new entry in the QA cache comprises the extracted one or more features of the input question stored in association with the one or more generated candidate answers for the input question.

5. The method of claim 3, further comprising:

in response to determining that there is not a matching entry in the one or more entries of the QA cache and in response to determining that a selected entry in the QA cache has a predetermined degree of similarity to the input question, updating the selected entry with the extracted one or more features of the input question in association with the one or more generated candidate answers for the input question.

6. The method of claim 1, wherein, in response to determining that there is not a matching entry in the one or more entries of the QA cache:

determining if there is a subset of entries in the one or more entries that is similar to the input question; and

outputting a listing of the subset of entries to a user for user selection of an entry in the subset of entries to be retrieved and used to generate candidate answers for the input question.

7. The method of claim 1, wherein entries in the QA cache comprise extracted features for previously processed questions with corresponding candidate answer infatuation and confidence measures associated with the candidate answers.

8. The method of claim 7, wherein the entries in the QA cache further comprise at least one of ranking information for the candidate answer, evidence information indicating the evidence used to generate the candidate answer, a determined domain of a corresponding previously processed question, information about a submitter of the corresponding previously processed question, or time of receipt of the corresponding previously processed question.