No-Fail Search System Using Dynamic Field Ranking

The patent segments the search fields into multiple independent components that can be individually removed. Instead of treating the keyword set as a single unit, the system divides it into separate search fields (e.g., product name, manufacturer, jurisdiction) and systematically removes the lowest-ranked ones to broaden the search while maintaining control over the broadening process.

The patent implements a dynamic search field removal process where the system automatically adjusts the set of search fields based on search performance. The system ranks search fields by productivity and dynamically removes the lowest-ranked field when a content type has zero results, creating an adaptive search strategy that responds to actual search outcomes.

The system dynamically adjusts the breadth of the search by removing search fields only when necessary (when a content type has zero results). This dynamic approach prevents unnecessary broadening that would reduce result quality, while still enabling broadening when needed to achieve the desired quantity of results.

The patent changes the parameters of the search by systematically removing search fields in a controlled manner. Instead of arbitrarily broadening the search, the system changes the search parameters by removing the lowest-ranked search field based on productivity metrics, ensuring that broadening occurs in a structured and optimized way.

The patent implements a self-service search system that automatically performs keyword adjustment and search broadening without user intervention. The system ranks search fields by productivity, automatically removes the lowest-ranked field when needed, and re-runs searches autonomously until satisfactory results are achieved across all content types, freeing the user from manual adjustment tasks.

The system uses feedback from search results to guide subsequent search adjustments. By organizing results according to content types and checking whether each content type has sufficient results, the system receives feedback on search performance and uses this feedback to determine which search field to remove next, creating a closed-loop search optimization process.

The patent segments the search results by content type (e.g., regulations, case law, verdicts, settlements) and independently evaluates the performance of the search for each content type. This segmentation allows the system to identify which specific content types are failing to return results and target the broadening process toward those specific areas.

The system dynamically adapts the search strategy for different content types by tracking the productivity of search fields across content types. When a specific content type has zero results, the system adjusts the search fields specifically for that content type by removing the lowest-ranked field, enabling adaptive optimization for each content type's unique characteristics.