Question Answering System Image Selection via Confidence Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Question answering systems often fail to provide comprehensive results, requiring users to perform separate searches for textual and image information, leading to inefficiencies and increased search time.
Innovation Solution
A deep question answering system that formulates answers to input questions, identifies relevant candidate images, and selects a subset based on confidence values and display specifications to provide content-relevant images alongside textual answers, optimizing search efficiency by integrating image and text searches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a question answering system provides only textual answers, then the system is simple to implement, but users must perform separate searches for image information, increasing search time
Solution Approach 1:
The patent combines text-based question answering with image search capabilities into a single integrated system. When a user asks a question, the system simultaneously retrieves textual answers and identifies relevant images from a corpus, presenting both together in one response. This eliminates the need for users to perform separate text and image searches, directly improving search efficiency while reducing total search time.
2Measurement precision
If the system identifies and displays multiple candidate images, then image relevance to the answer is improved, but the complexity of image selection and processing increases
Solution Approach 1:
The system employs a confidence value mechanism that provides feedback on the relevance of each candidate image to the generated answer. Images are scored based on their relationship to the answer content, and this confidence metric guides the selection process. The system can display top-ranked images with highest confidence values, or adjust display based on threshold confidence levels, thereby managing complexity while maintaining high image relevance.
3Loss of information
If the system displays all candidate images, then completeness of information is improved, but the display area constraints and user interface complexity increase
Solution Approach 1:
Instead of displaying all candidate images, the system selectively displays a subset based on confidence values and display specifications. It identifies and presents only the most relevant images that fit within display constraints, rather than attempting to show every possible image. This partial action approach maintains information completeness for the most important content while respecting display area limitations and reducing interface complexity.
Data Source
AI summary
A method for providing content relevant images for an input question to a deep question answering system is disclosed. The method can include formulating, in response to receiving the input question, an answer to the input question. The method can also include identifying, based on the answer to the input question, a set of candidate images from a corpus of information. The set of candidate images may relate to the answer. The method can also include determining, based on a confidence value of the set of candidate images and a designated display specification, a first subset of the set of candidate images to provide in a display area. The method can also include selecting the first subset to be provided in the display area.