Contextual Variable Selection via Similarity Graph
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional visualization computing systems provide poorly organized variable selection menus, leading to a burdensome user experience as users navigate through long lists of irrelevant variables to select relevant data set variables for generating visual representations.
Innovation Solution
The system generates a variable similarity graph based on computed feature vectors, allowing for the creation of contextual menus that prioritize relevant variables, reducing the need for extensive navigation by promoting more relevant variables and hiding less relevant ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If variables are listed in alphabetical or numerical order in the selection menu, then all variables can be displayed systematically, but users must scroll through many irrelevant variables to find relevant ones
Solution Approach 1:
The system pre-computes feature vectors and similarity metrics for all variables before the user needs to select them. This preliminary analysis enables the menu to be dynamically organized based on relevance to the currently selected variable, rather than requiring users to manually search through alphabetically or numerically sorted lists.
Solution Approach 2:
The variable selection menu is organized differently depending on the context - specifically, based on the currently selected variable. Variables are grouped and ranked according to their similarity to the selected variable, creating a localized, context-aware ordering that adapts to user needs rather than using a single global sorting method.
2Ease of operation
If a contextual menu with similarity-based variable recommendations is implemented, then relevant variables are prioritized and accessibility is improved, but the system complexity increases due to graph computation
Solution Approach 1:
The system automatically computes feature vectors, builds similarity graphs, and generates contextual variable recommendations without requiring user intervention. The computation is performed autonomously based on the data characteristics, and the results are seamlessly integrated into the user interface, making the complexity transparent to the user.
Solution Approach 2:
The patent replaces manual variable selection (mechanical searching through lists) with an automated information retrieval system based on computational similarity analysis. Instead of users manually navigating through sorted lists, the system computationally determines variable relationships and presents them in an optimized order.
Data Source
AI summary
This disclosure involves generating, from a user data set, a ranked list of recommended secondary variables in a user interface field similar to primary variable selected in another user interface field. A system receives a data set having variables and corresponding sets of values. The data visualization system determines a feature vector for each variable based on statistics of a corresponding values set. The system generates a variable similarity graph having nodes representing variables and links representing degrees of similarity between feature vectors of variables. The system receives a selection of a first variable via a first field of the user interface, detects a selection of a second field, and identifies a relationship between the first field and the second field. The system generates a contextual menu of recommended secondary variables for use with the selected first variable based on similarity value of the links in the variable similarity graph.