Back to results
US 10,540,411 B2Granted
Candidate answers for speculative questions in a deep question answering system
AssigneeInternational Business Machines Corporation
InventorsAdam T. Clark, Jeffrey K. Huebert, John E. Petri
Life of this patent
Estimating the term…
Abstract
System, method, and computer program product to determine that a question received by a deep question answering system is speculative, generate, by one or more predictive algorithms, a set of candidate answers, compute a score for each candidate answer in the set of candidate answers, and return a first candidate answer, of the set of candidate answers, as responsive to the speculative question received by the deep question answering system.
Claims10
A computer-implemented method, comprising:
- analyzing a text of a question received by a question answering system executing on a processor, wherein the question answering system uses a corpus of stored information comprising a plurality of electronic documents to generate answers, and wherein analyzing the text comprises identifying a future tense in the question;
- determining, based on the analysis of the text of the question, that an answer to the question is not explicitly contained in the corpus of stored information;
- upon determining that an answer to the question is not explicitly contained in the corpus, determining that the plurality of electronic documents in the corpus does not include a candidate answer for the question;
- and
- upon determining that the plurality of electronic documents in the corpus does not include a candidate answer for the question:
- determining that the question is speculative and cannot be answered definitively at a current time;
- identifying, based on the analysis, an attribute of the question, wherein the attribute of the question relates to a desired response to the question;
- generating, by one or more predictive algorithms executing on the processor, a plurality of candidate answers for the question based on the identified attribute of the question, wherein each of the one or more predictive algorithms are used to generate estimated future values for the attribute of the question;
- computing a respective score for each candidate answer of the plurality of candidate answers;
- and
- returning, by the question answering system executing on the processor, a first candidate answer, of the plurality of candidate answers, as responsive to the speculative question received by the question answering system.
8 more elements
A computer-implemented method, comprising:
- analyzing a text of a question received by a question answering system executing on a processor, wherein the question answering system uses a corpus of stored information comprising a plurality of electronic documents to generate answers, and wherein analyzing the text comprises identifying:
- (i) a future tense in the text of the question, and (ii) a concept corresponding to an independent variable in the question;
- determining, based on the identified future tense and concept corresponding to the independent variable, that an answer to the question is not explicitly contained in the corpus of information;
- upon determining that an answer to the question is not explicitly contained in the corpus, determining that the corpus does not include a candidate answer for the question;
- and
- upon determining that the corpus does not include a candidate answer for the question:
- determining that the question is speculative and cannot be answered definitively at a current time;
- identifying, based on the analysis, an attribute of the question, wherein the attribute of the question relates to a desired response to the question;
- upon determining that a predictive algorithm related to the independent variable does not exist, generating a first predictive algorithm based on the identified attribute of the question, wherein the first predictive algorithm is used to generate estimated future values for the attribute of the question;
- generating, by the first predictive algorithm executing on the processor, a plurality of candidate answers for the question based on the attribute;
- computing a respective score for each candidate answer of the plurality of candidate answers;
- and
- returning, by the question answering system executing on the processor, a first candidate answer, of the plurality of candidate answers, as responsive to the question.
10 more elements
Dependencies are inferred from the claim text.