Enhanced answers in DeepQA system according to user preferences
A semantic search engine is enhanced to employ user preferences to customize answer output by, for a first user, extracting user preferences and sentiment levels associated with a first question; receiving candidate answer results of a semantic search of the first question; weighting the candidate answer results according to the sentiment levels for each of the user preferences; and producing the selected candidate answers to the first user. Optionally, user preferences and sentiment levels may be accumulated over different questions for the same user, or over different users for similar questions. And, supplemental information may be retrieved relative to a user preference in order to further tune the weighting per the preferences and sentiment levels.
1. A method for using user preferences to customize answer output comprising:
extracting, by a natural language processor computer, from a first question, one or more preferences for a first user;
extracting, by the natural language processor computer, from the first question, one or more sentiment levels for the first user;
performing by a computer a semantic search on the first question;
receiving by a computer from the semantic search, a first plurality of candidate answers to the first question;
score, using the preferences and the sentiment levels extracted from the first question, each candidate answer to the first question;
selecting by a computer one or more of the received candidate answer to the first question according to each corresponding score;
outputting by a computer to the first user, via a user interface device, the selected one or more candidate answers to the first question,
wherein the first user interface device comprises an output device selected from the group of a display, a printout, and a speaker;
extracting, by the natural language processor computer, from a second question, one or more preferences for the first user;
extracting, by the natural language processor computer, from the second question, one or more sentiment levels for the first user;
accumulating the preferences extracted from the second question, the sentiment levels extracted from the second question, and the preferences extracted from the first question, and the sentiment levels extracted from the first question, to form a set of accumulated preferences and sentiment levels;
performing by a computer a second semantic search on the second question;
receiving from the second semantic search a plurality of candidate answers to the second question;
scoring by a computer, using the set of accumulated preferences and sentiment levels, each candidate answer to the second question;
selecting by a computer one or more of the received candidate answers to the second question according to each corresponding score;
outputting by a computer, to the first user, via a user interface device, the selected one or more candidate answers to the second question, wherein the user interface device comprises an output device selected from the group consisting of a display, a printout, and a speaker.
2. The method as set forth in claim 1 further comprising:
extracting, by the natural language processor computer, from a third question, one or more preferences for a second user;
extracting, by the natural language processor computer, from the third question, one or more sentiment levels for the second user;
accumulating, by a computer, the preferences extracted from the third question and the sentiment levels extracted from the third question into the set of accumulated preferences and sentiment levels;
performing, by a computer, a semantic search on the third question;
receiving, by a computer, from the semantic search, a plurality of candidate answers to the third question;
scoring, by a computer, using the set of accumulated preferences and sentiment levels, each candidate answer to the third question;
selecting, by a computer, one or more of the received candidate answers to the third question according to each corresponding score;
outputting to the second user, via a user interface device, the selected one or more candidate answers to the third question, wherein the user interface device comprises an output device selected from the group consisting of a display, a printout, and a speaker.
3. The method as set forth in claim 2 further comprising detecting, by a computer, that the first question is similar to at least the second question.
4. The method as set forth in claim 1 further comprising retrieving, by a computer, supplemental information regarding at least one of the candidate answers, wherein the supplemental information is related to one or more of the user preferences, and wherein the supplemental information is used to adjust the selecting.
5. The method as set forth in claim 1 further comprising extracting, by the natural language processor computer, semantic elements according to a standardized architecture for managing unstructured information.
6. The method as set forth in claim 5 wherein the standardized architecture for managing unstructured information conforms to Unstructured Information Management Architecture (UIMA).
7. The method as set forth in claim 1 wherein the selecting of candidate answers comprises a performing a weighting method.