Browse Prior Art Database

System and Method for Refined (Cognitive) Sets of Questions and Answers Disclosure Number: IPCOM000247474D
Publication Date: 2016-Sep-09
Document File: 4 page(s) / 66K

Publishing Venue

The Prior Art Database


Disclosed is a method/process that is applicable to a cognitive Question-and-Answer system, which analyzes and reformulates user questions into more effective queries for a cognitive system. This allows the cognitive analysis system to return cognitive matches/answers with high confidence.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 42% of the total text.

Page 01 of 4

System and Method for Refined (Cognitive) Sets of Questions and Answers

An existing cognitive Question-and-Answer (QA) system understands complex questions; it evaluates all possible meanings, determines the context of the question, and then presents relevant answers and solutions based on supporting evidence and the quality of the information found. The response is a pool of candidate answers, with associated confidence levels and links to supporting evidence. Cognitive systems put content into context, providing confidence-weighted responses with supporting evidence.

A QA system, however, does not always capture the information for which the user is searching. The question-processing component might not properly classify the question or the information needed for extracting and generating the answer; therefore, the correct answer is not easily retrieved. In such cases, the user might not only want to reformulate the question, but also have a better question-answer system/method to use.

A method is needed to automatically improve the user's text entry in a cognitive QA system in order to provide relevant results.

The novel contribution is a method/process that enhances the accuracy of the user's text entry in a cognitive QA system. The method transforms the entry into better cognitive questions that provide more sets of cognitive answers with improved relevancy to the first initial question.

Through the usage of a cognitive system's message resonances service analysis, the novel method creates multiple candidate cognitive questions. This produces more cognitive and better answers. The system transforms the existing QA service to a cognitive (selectively, persistent and better) conversation. The user enters a question (typically by typing). The QA system accepts the question as input, and then generates and displays one or more responses, as well as reformulated questions and cognitive responses.

The proposed system addresses the user interaction with a QA system in which multiple question-response sets are generated. The method allows the conversation thread to be displayed in the user interface as additional questions are asked.

The main components and implementation process follow:

1. A system that automatically processes the Question/Answer(s) pairs A. For questions posted by a user (in a natural language) as a Request: i. server application transforms it to a JavaScript* Object Notation (JSON) value and sends it to the cognitive system's QA Application Programming Interface (API) service as a Request JSON

ii. system receives the cognitive component's QA API Service

Response JSON and includes (pass throughout) the user's browser application, which it uses to create the output format (i.e., the answers part)


Page 02 of 4

B. The system processes the user's entry
i. analyzes the user's question
ii. creates a candidate cognitive question
iii. sends the cognitive question to the cognitive system's Q/A API
iv. receives the answers