Collaborative Filtering for Interactive Visualization Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of business analytics applications face challenges in navigating interactive visualizations, as the vast number of possible views makes it difficult to determine which ones contain insightful information, with current systems providing little support for finding helpful views.
Innovation Solution
Implementing collaborative filtering methods that generate inquiry histories based on user interactions with interactive visualizations, allowing for recommendations on next actions by tracking user interactions, current states, and navigation patterns to suggest useful views or paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the visualization allows users to explore all possible views by manipulating parameters (region, product, time, etc.), then the completeness of data exploration is improved, but the complexity of navigation and the time required to find insightful views increases significantly
Solution Approach 1:
The system performs preliminary actions by pre-computing and storing inquiry histories from multiple users before the current user needs to explore. These pre-collected navigation patterns and viewed views are stored in a database, ready to be quickly retrieved and presented as recommendations, eliminating the need for the user to manually explore all possible views
Solution Approach 2:
The patent introduces an intermediary recommendation system that mediates between the vast space of possible visualization views and the user. This intermediary analyzes inquiry histories, determines similarity between current and historical inquiry states, and presents filtered recommendations, thereby reducing the direct navigation burden on the user while maintaining access to comprehensive data exploration options
2Ease of operation
If the system provides detailed navigation support and recommendations for all possible views, then the ease of finding insightful views is improved, but the computational complexity and data processing requirements increase
Solution Approach 1:
The system applies local quality by providing recommendations selectively rather than uniformly for all possible views. It calculates similarity between the current inquiry state and historical states, then provides recommendations only for views that are locally relevant (high similarity) to the user's current context, rather than processing and presenting all possible views equally
Solution Approach 2:
The patent changes parameters by transforming the raw inquiry history data into similarity scores and ranked recommendations. Instead of processing all possible view combinations, it modifies the data representation to focus on key distinguishing features of inquiry states, enabling efficient comparison and recommendation generation with reduced computational complexity
3Measurement precision
If the system tracks and stores detailed inquiry histories for all users, then the accuracy of personalized recommendations is improved, but the quantity of data to be processed and stored increases
Solution Approach 1:
The system extracts only the essential and relevant features from detailed inquiry histories for recommendation purposes. Instead of processing and storing all raw interaction data, it identifies and extracts key elements such as inquiry state characteristics, navigation patterns, and view preferences that are most predictive of useful recommendations, discarding redundant information
Data Source
AI summary
Embodiments of the invention provide systems and methods for navigating interactive visualizations of a business analysis application based on collaborative filtering. More specifically, embodiments of the present invention provide a recommender that functions together with a visualization tool and business analytics application. This recommender can track use of interactive visualizations provided by the visualization tool, e.g., views selected, functions performed, navigation between views, etc., by various users to build a set of inquiry histories. Then, based on these histories and possibly other considerations, recommendations can be made to a current user as to which views, functions, etc. might be useful or insightful. In other words, embodiments of the present invention track the analysis behavior of each user and recommend which views may be of interest for the corresponding analysis task based on the behavior of similar users in similar situations.