Spreadsheet Rule Generation From Formatting Examples
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users often find it tedious and error-prone to manually create data processing rules in spreadsheet programs, particularly for conditional formatting, due to the complexity of rule syntax and data logic, leading to inefficient and incorrect formatting.
Innovation Solution
A machine learning architecture, such as CORNET, automatically generates data processing rules, including conditional formatting rules, by analyzing user examples of formatted cells, utilizing semi-supervised clustering and neural ranking to suggest simplified rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If users manually create data processing rules in spreadsheet programs, then they can achieve precise control over formatting logic, but the process becomes tedious and error-prone due to rule syntax complexity
Solution Approach 1:
The patent introduces an intermediary system (rule generation engine with machine learning models) that mediates between user intent and complex rule syntax. Users provide simple criteria or examples, and the system automatically generates syntactically correct formatting rules, eliminating the need for users to directly manipulate complex rule syntax while maintaining precise control over formatting logic.
Solution Approach 2:
The system enables self-service by allowing users to create formatting rules through intuitive interactions such as selecting cells or providing examples, without requiring manual syntax construction. The rule generation engine automatically processes user input and generates complete formatting rules, making the system serve itself in translating simple user actions into complex rule formulations.
2Adaptability or versatility
If users manually create data processing rules, then they can handle complex formatting scenarios, but the process becomes time-consuming and inefficient
Solution Approach 1:
The system performs preliminary action by pre-processing user input and automatically generating rule structures before final rule application. The rule generation engine prepares formatting rules in advance based on user criteria or examples, so when users need formatting applied, the rules are already formulated and ready for immediate use, significantly reducing creation time.
Solution Approach 2:
The patent replaces the mechanical system of manual rule construction with an automated intelligent system. Instead of users manually assembling rule components through complex syntax operations, a machine learning-based engine automatically generates complete formatting rules, substituting human manual effort with automated computational processes that are both faster and more capable.
3Ease of operation
If traditional rule generation methods are used, then users have full control over rule creation, but the process requires significant user expertise and effort
Solution Approach 1:
The intermediary rule generation engine shields users from system complexity by handling all complex rule syntax and logic internally. Users interact with a simplified interface that requires no expertise in formatting rules, while the engine manages the complex transformations, maintaining user control through intuitive input options without exposing underlying complexity.
Solution Approach 2:
The system uses copying by analyzing user-provided examples or selected cells and replicating their formatting patterns automatically. Instead of requiring users to understand and construct complex rules from scratch, the system copies existing formatting patterns and generalizes them into reusable rules, reducing the perceived complexity while maintaining full user control over the output.
Data Source
AI summary
Some embodiments automatically generate data processing rules based on positive examples of processed data, e.g., formatting rules based on formatted data, filtering rules based on filtered data, or validating rules based on valid data. Some embodiments also use negative examples, e.g., unformatted data. A machine learning rule generation architecture includes a predicate generator, a cell cluster creator, a rule enumerator, and in some versions a rule ranker. Formatting rules written by a user are replaced by simpler autogenerated rules. Spreadsheet formatting rule functionality is enhanced, and surfaced in a user interface.