Sequence-to-Sequence Query Rewriting for Natural Language Intent
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current information retrieval systems face challenges in processing natural language queries due to their fuzziness and implicitness, often requiring multiple user interactions for clarification, as they struggle to capture the intent behind such queries.
Innovation Solution
A method using a sequence-to-sequence model with an attention layer to convert natural language queries into standard queries, allowing for more accurate document retrieval by scoring documents based on conditional entropy, and generating follow-up questions to enhance relevance when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If natural language queries are simplified by removing stop words, then key terms remain for retrieval, but the intent of the query is not captured
Solution Approach 1:
The patent introduces an intermediary component (query rewriting system with sequence-to-sequence model) that transforms the original natural language query into a rewritten query that better captures user intent. This intermediary processing layer preserves semantic meaning while generating queries suitable for information retrieval systems, resolving the conflict between simplification and intent preservation.
Solution Approach 2:
The system changes the parameter of query representation from simple keyword extraction to transformed natural language queries generated by neural network models. This parameter change allows the system to maintain both the simplicity needed for processing and the richness needed to capture intent, by dynamically generating optimized query formulations.
2Productivity
If existing information retrieval systems process natural language queries directly, then multiple user interactions are required for clarification, but processing efficiency is reduced
Solution Approach 1:
The system performs preliminary action by pre-processing and rewriting queries before they are submitted to the information retrieval system. The sequence-to-sequence model generates optimized queries in advance, capturing user intent upfront and reducing the need for subsequent clarification interactions, thus improving overall efficiency and reducing user time investment.
3Reliability
If standard keyword-based search is used, then search execution is fast, but results do not correspond to user intent
Solution Approach 1:
The patent replaces the mechanical keyword-matching system with a neural network-based sequence-to-sequence model that understands and transforms natural language queries. This substitution enables the system to capture semantic intent and generate more relevant search queries, improving result reliability while maintaining processing efficiency through automated query transformation.
Data Source
AI summary
The present application describes a system and method for converting a natural language query to a standard query using a sequence-to-sequence neural network. As described herein, when a natural language query is receive, the natural language query is converted to a standard query using a sequence-to-sequence model. In some cases, the sequence-to-sequence model is associated with an attention layer. A search using the standard query is performed and various documents may be returned. The documents that result from the search are scored based, at least in part, on a determined conditional entropy of the document. The conditional entropy is determined using the natural language query and the document.