Cognitive question answering pipeline blending
AssigneeInternational Business Machines Corporation
InventorJohn M. Boyer
Life of this patent
Estimating the term…
Abstract
An answer to a question may selected from answers from a set of answering pipelines. Question answer data can be generated for a question, using a first answering pipeline. Another set of question answer data can be generated for the second question, using the second answering pipeline. The question answer data can include answers and confidence values for each answer. Using a weighting formula and a blending profile for the first answering pipeline, a vote weight can be determined for an answer with the highest confidence value. The same weighting formula and a second blending profile may be used to determine a vote weight for another answer with the highest confidence value. An answer to the question may be selected from the answers, based on the overall highest vote weight.
Claims20
A method comprising:
- generating, for a question, a first set of question answer data using a first answering pipeline, wherein the first set of question answer data comprises a first set of answers and a set of first pipeline confidence values for each answer in the first set of answers;
- generating, for the question, a second set of question answer data using a second answering pipeline, wherein the second set of question answer data comprises a second set of answers and a set of second pipeline confidence values for each answer in the second set of answers;
- determining, using a weighting formula and a first blending profile associated with the first answering pipeline, a first vote weight for an answer in the first set of question answer data that was assigned a first pipeline highest confidence value by the first answering pipeline;
- determining, using the weighting formula and a second blending profile associated with the second answering pipeline, a second vote weight for an answer in the second set of question answer data that was assigned a second pipeline highest confidence value by the second answering pipeline;
- and
- selecting, as a first answer to the question, the answer with an overall highest vote weight from among the answers in the first set of question answer data and in the second set of question answer data.
A system comprising:
- a computer readable storage medium with program instructions stored thereon;
- and
- one or more processors configured to execute the program instructions to perform a method comprising:
- generating, for a question, a first set of question answer data using a first answering pipeline, wherein the first set of question answer data comprises a first set of answers and a set of first pipeline confidence values for each answer in the first set of answers;
- generating, for the question, a second set of question answer data using a second answering pipeline, wherein the second set of question answer data comprises a second set of answers and a set of second pipeline confidence values for each answer in the second set of answers;
- determining, using a weighting formula and a first blending profile associated with the first answering pipeline, a first vote weight for an answer in the first set of question answer data that was assigned a first pipeline highest confidence value by the first answering pipeline;
- determining, using the weighting formula and a second blending profile associated with the second answering pipeline, a second vote weight for an answer in the second set of question answer data that was assigned a second pipeline highest confidence value by the second answering pipeline;
- and
- selecting, as a first answer to the question, the answer with an overall highest vote weight from among the answers in the first set of question answer data and in the second set of question answer data.
A computer program product comprising a computer readable storage medium having program instructions embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, the program instructions executable by a computer processing circuit to cause the circuit to perform the method comprising:
- generating, for a question, a first set of question answer data using a first answering pipeline, wherein the first set of question answer data comprises a first set of answers and a set of first pipeline confidence values for each answer in the first set of answers;
- generating, for the question, a second set of question answer data using a second answering pipeline, wherein the second set of question answer data comprises a second set of answers and a set of second pipeline confidence values for each answer in the second set of answers;
- determining, using a weighting formula and a first blending profile associated with the first answering pipeline, a first vote weight for an answer in the first set of question answer data that was assigned a first pipeline highest confidence value by the first answering pipeline;
- determining, using the weighting formula and a second blending profile associated with the second answering pipeline, a second vote weight for an answer in the second set of question answer data that was assigned a second pipeline highest confidence value by the second answering pipeline;
- and selecting, as a first answer to the question, the answer with an overall highest vote weight from among the answers in the first set of question answer data and in the second set of question answer data.
Dependencies are inferred from the claim text.