Chart Recommendation Vector Similarity Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional chart recommendation methods based on pre-defined decision trees are inadequate for scenarios with many types of charts and complex fields, leading to inaccurate and complex recommendations, with high maintenance costs and limited ability to handle new chart types or data.

Innovation Solution

A method that generates an input vector for each chart based on user input fields, calculates similarity with predetermined feature vectors, and recommends charts based on these similarities, reducing complexity and improving accuracy by dynamically adjusting input field purposes, types, and units.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If conventional decision tree methods are used for chart recommendation, then the system structure is simple and easy to implement, but the recommendation accuracy deteriorates in scenarios with many chart types and complex fields

Engineering Contradiction:

Improvesystem structureVSAvoidrecommendation accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent replaces the mechanical decision tree system with an AI-based recommendation system that uses neural networks and deep learning models to automatically learn chart selection patterns from data, achieving higher accuracy without requiring complex manual rule construction

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system dynamically adjusts recommendation parameters based on data characteristics, chart type, and user behavior patterns, allowing the recommendation accuracy to adapt to different scenarios with many chart types and complex fields

Inventive Principle:

Principle #35Parameter changes

2Ease of manufacture

If pre-defined decision trees are used, then the implementation is straightforward, but the ability to handle new chart types or data deteriorates

Engineering Contradiction:

Improveimplementation easeVSAvoidhandling new chart types

Core Design Contradiction:

Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The recommendation system is designed to be dynamic and adaptive, automatically learning from new data and chart types as they emerge, rather than relying on static pre-defined decision trees that require manual updates when new chart types are introduced

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system performs self-learning and self-optimization by analyzing user interactions and data patterns, enabling it to automatically adapt to new chart types and data structures without requiring manual reconfiguration of the decision tree

Inventive Principle:

Principle #25Self-service

3Ease of repair

If decision tree-based recommendation is used, then the system is easy to maintain, but the maintenance cost increases when new chart types or data are added

Engineering Contradiction:

Improvemaintenance easeVSAvoidmaintenance cost

Core Design Contradiction:

Ease of repairVSQuantity of substance

Solution Approach 1:

The patent replaces manual decision tree maintenance with automated AI-based learning mechanisms that continuously optimize recommendations based on actual usage data, eliminating the need for manual updates when new chart types are added

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11630827B2Method for recommending chart, electronic device, and storage medium

Publication Date: 2023.04.18 BEIJING BAIDU NETCOM SCI & TECH CO LTD

AI summary

The disclosure provides a method and an apparatus for recommending a chart, an electronic device, and a storage medium. The method may include: generating an input vector of at least one input field relative to each chart based on the at least one input field obtained in advance; calculating a similarity of the input vector and a predetermined feature vector corresponding to each chart; obtaining a target chart corresponding to the at least one input field based on the similarity between the input vector and the predetermined feature vector corresponding to each chart; and sending the target chart to a terminal device.