Question Answering System Using Integrated Comment Representations

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing intelligent question and answer robots are limited in responding to user questions as they primarily focus on specific dimensions and struggle to cover other relevant aspects, making it difficult to provide accurate answers.

Innovation Solution

A method and apparatus that determine question and comment characteristic representations, integrate them to generate a target characteristic representation, and use this representation to accurately answer user questions, considering multiple dimensions and comments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If intelligent question and answer robots focus on specific dimensions, then they can provide focused responses, but they fail to cover other relevant aspects and provide accurate answers

Engineering Contradiction:
Improveresponse accuracyVSAvoiddimension coverage
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system integrates multiple comment representations (text, image, audio) and multiple question processing approaches into a unified answer generation framework, enabling the Q&A robot to handle diverse question types and dimensions simultaneously, thus achieving both focus and comprehensiveness

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent combines question characteristic representation with multiple comment characteristic representations to generate integrated target representations, merging information from different sources and dimensions to improve both accuracy and coverage of responses

Inventive Principle:
Principle #5Merging (Combining)

2Measurement precision

If the system integrates multiple comment representations and dimensions, then it improves answer accuracy and comprehensiveness, but it increases computational complexity

Engineering Contradiction:
Improveanswer accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the complex processing task into distinct modules: question representation module, comment representation module, integration module, and answer generation module. Each module handles specific processing tasks independently, making the overall complex system more manageable and efficient

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces characteristic representations as intermediary structures that bridge questions and comments. These representations serve as intermediate processing forms that simplify the integration of heterogeneous data types and reduce the complexity of direct multi-dimensional analysis

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS11461399B2Method and apparatus for responding to question, and storage medium
Publication Date: 2022.10.04 BAIDU ONLINE NETWORK TECH (BEIJIBG) CO LTD
  • US11461399B2 patent drawing
  • US11461399B2 patent drawing
  • US11461399B2 patent drawing

AI summary

According to embodiments of the present disclosure, a method and an apparatus for responding to a question, and a storage medium are provided. The method includes: determining a question characteristic representation corresponding to a question for an object; determining a comment characteristic representation corresponding to a first comment for the object; generating a first target characteristic representation by utilizing the question characteristic representation and the comment characteristic representation; and determining an answer for the question based on the first target characteristic representation.