Neural Query Correction and Completion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search query systems face inefficiencies in correcting and completing misspelled queries, requiring a two-step process that can be computationally inefficient and struggle to accurately rank relevance of suggestions, especially when the misspelled query has never been previously entered.
Innovation Solution
A machine-learning model using a deep neural network encoder-decoder architecture is trained to simultaneously generate spelling-corrected and completed query suggestions, learning from past queries to generalize corrections and completions, including common misspellings due to typographical errors and phonetic variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a two-step process is used to correct and complete search queries, then the system can handle misspelled queries, but the computational efficiency decreases and the process becomes more complex
Solution Approach 1:
The patent combines the query correction and query completion tasks into a single unified neural network model. The encoder-decoder architecture processes the input query (whether misspelled or not) and directly generates completed query suggestions in one step, eliminating the need for separate correction and completion steps. This merging of functions improves computational efficiency while maintaining accuracy.
2Productivity
If traditional query completion is used without correction, then the process is simpler and faster, but the accuracy of suggested results decreases when queries are misspelled
Solution Approach 1:
The neural network model performs correction action preliminarily within its processing pipeline. When the model receives a misspelled query, it internally identifies and corrects the spelling errors as part of its encoding process, before generating completion suggestions. This preliminary correction action ensures accurate results while maintaining efficient single-step processing.
3Ease of manufacture
If separate correction and completion steps are implemented, then each step can be optimized independently, but the overall system complexity increases
Solution Approach 1:
The neural network model is designed as a universal system that handles both correction and completion functions simultaneously. The encoder-decoder architecture is multi-functional: the encoder processes various types of input queries (misspelled or correct), and the decoder generates appropriate completions. This universal design eliminates the need for separate specialized components, reducing overall system complexity while maintaining ease of implementation.
Data Source
AI summary
Aspects of subject technology provide systems and methods for simultaneously spell-correcting and completing partial search queries being entered by a user on the user's electronic device. An apparatus such as a computing device may receive partial search queries from the user's electronic device as each character of the partial search query is entered by the user. The apparatus may utilize a machine-learning model to generate suggested queries that include spelling-corrected versions of the received partial query, query completion suggestions for the partial query, and/or spelling-corrected completion suggestions for the partial query.