Hierarchical Item Graph Query Reformulation for Search Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current online concierge systems struggle to efficiently match customer search queries with available items due to mismapped terms and incomplete item graph representations, leading to suboptimal item recommendations.
Innovation Solution
An online concierge system utilizes an item graph with hierarchical taxonomy, segments search queries into tokens, and applies a query reformulation model to broaden search terms, enhancing item matching and recommendation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the online concierge system uses exact search term matching in the item graph, then search precision is improved, but search recall deteriorates when terms do not map to available items
Solution Approach 1:
The system dynamically changes the parameter of search term granularity by reformulating queries to map to different hierarchical levels in the item graph. When exact matches fail, the system transitions to broader category levels, adjusting the precision-recall balance based on item availability at each hierarchical level.
Solution Approach 2:
The patent introduces an intermediary reformulation process that acts as a mediator between the customer's search query and the item graph. This intermediary layer translates unmatched search terms into appropriate hierarchical node mappings, enabling the system to bridge the gap between customer intent and available inventory.
2Measurement precision
If the system provides detailed specific item results, then recommendation precision is improved, but the system fails to provide results when items are unavailable
Solution Approach 1:
Instead of starting from the customer's specific search term and drilling down to match items, the system inverts the approach by starting from available items at higher hierarchical levels and working downward. When specific items are unavailable, the system retrieves items from broader categories, ensuring result availability while maintaining relevance.
Solution Approach 2:
The system performs preliminary checks at higher hierarchical levels before committing to specific item recommendations. By pre-evaluating item availability at parent category levels, the system ensures that recommendations can be generated even when specific sub-category items are out of stock.
3Measurement precision
If the item graph maintains detailed hierarchical taxonomy, then item categorization accuracy is improved, but query matching complexity increases
Solution Approach 1:
The patent segments the query matching process into distinct hierarchical stages corresponding to different levels of the item graph. Rather than attempting to match queries against the entire taxonomy simultaneously, the system processes matches level-by-level from root to leaves, reducing computational complexity while preserving categorization accuracy.
Data Source
AI summary
An online concierge system generates an item graph connecting item nodes with attribute nodes of the items. When the online concierge system receives a search query to identify one or more items from a customer, the online concierge system parses the search query into combinations of terms and identifies item nodes and attribute nodes related to the search query. The online concierge system may determine that no item nodes meet presentation criteria. The online concierge system may determine that a reformulated search query has a higher conversion probability than the search query received from the customer. The online concierge system reformulates the search query. The online concierge system selects item nodes as search results. The online concierge system transmits the search results to the customer.