Causal Language Model Query Expansion for Image Diversity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional query processing systems are unable to automatically generate alternative or augmented queries based on an original user query, resulting in a limited and static number of images provided to users.
Innovation Solution
A query processing apparatus that utilizes a causal language model (CLM) to generate expanded queries by adding additional phrases and a masked language model (MLM) to replace mask tokens with insertion phrases, providing a diverse set of relevant images by combining the 'broaden intent' and 'narrow intent' phases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional query processing systems are used, then the system structure is simple, but the diversity of output images is limited
Solution Approach 1:
The query processing system is segmented into multiple specialized language models (causal language model for query expansion, masked language model for phrase replacement) that each handle specific aspects of query transformation, enabling diverse image output through coordinated specialized components
Solution Approach 2:
Language models serve as intermediary components between the user's original query and the image generation system, transforming and expanding queries into multiple alternative forms that broaden the scope of generated images without requiring direct modification of the image generation core
2Adaptability or versatility
If query expansion is implemented using language models, then the diversity of images is improved, but the processing time increases
Solution Approach 1:
Language models pre-generate multiple expanded query variations and alternative phrases before image generation occurs, allowing the image generation system to work with ready-prepared diverse queries rather than generating them during the imaging process
Solution Approach 2:
The system generates a controlled set of expanded queries (excessive action) to ensure sufficient diversity for image generation, balancing the need for variety with acceptable processing time by limiting the expansion to a manageable number of alternatives
3Ease of operation
If multiple language models are used for query expansion, then the query processing capability is enhanced, but the device complexity increases
Solution Approach 1:
Multiple language models are integrated into a unified query processing apparatus that handles diverse query transformation tasks (expansion, replacement, rephrasing) through a single coordinated system, making the complex multi-model capability accessible through a simple user interface
Data Source
AI summary
Systems and methods for query processing are described. Embodiments of the present disclosure identify an original query; generate a plurality of expanded queries by generating a plurality of additional phrases based on the original query using a causal language model (CLM) and augmenting the original query with each of the plurality of additional phrases, respectively; and provide a plurality of images in response to the original query, wherein the plurality of images are associated with the plurality of expanded queries, respectively.