Neural Network Query Prediction System

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional search engines face significant challenges in providing timely results as dataset sizes grow, often requiring users to run modified queries that yield less relevant results within an acceptable time frame, failing to deliver timely and desired insights from historical transaction data.

Innovation Solution

A computerized method utilizing a neural network to predict user queries by analyzing historical search patterns, extracting query parameters, and generating predicted queries before user initiation, thereby pre-searching large datasets and providing immediate results.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If conventional search engines are used to query large datasets, then query completeness can be maintained, but query time increases significantly

Engineering Contradiction:

Improvequery timeVSAvoidresults timeliness

Core Design Contradiction:

Loss of timeVSProductivity

Solution Approach 1:

The system performs preliminary actions by predicting user queries using a neural network model trained on historical query data, and pre-executing these predicted queries before the user actually submits them. This allows results to be ready in advance, eliminating the need to wait for query execution when the user needs them.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system serves itself by automatically generating predicted queries through the neural network model and executing them without requiring user intervention. The model continuously learns from historical data and autonomously determines what queries to run, reducing manual effort and accelerating result delivery.

Inventive Principle:

Principle #25Self-service

2Loss of time

If users modify queries to reduce execution time, then query time decreases, but result relevance deteriorates

Engineering Contradiction:

Improvequery execution timeVSAvoidresult relevance

Core Design Contradiction:

Loss of timeVSLoss of information

Solution Approach 1:

The system pre-executes complete, relevant queries before users submit them, so when users need results, they immediately receive comprehensive answers without having to simplify their queries. The full query execution happens in the background, maintaining result quality while delivering speed.

Inventive Principle:

Principle #10Preliminary action

3Quantity of substance

If dataset size increases to store more historical data, then data completeness improves, but query performance deteriorates

Engineering Contradiction:

Improvedataset sizeVSAvoidquery speed

Core Design Contradiction:

Quantity of substanceVSSpeed

Solution Approach 1:

The neural network model analyzes historical query patterns from large datasets and pre-executes predicted queries, so that even as datasets grow, the system maintains fast response times by having results ready in advance based on learned patterns, rather than scanning entire datasets in real-time.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11042538B2Predicting queries using neural networks

Publication Date: 2021.06.22 MASTERCARD INT INC

AI summary

A system for generating queries accesses a query history for a user. The query history includes a plurality of queries having defined query parameters. The query parameters are extracted from the plurality of queries and input into a neural network. The neural network generates an output corresponding to a predicted query the output is used to generate a predicted query and run the predicted query to generate a query result. By running the predicted query prior to a user requesting the query, results are thereby provided without lengthy processing delays when the user requests the predicted query.