Search Query Relaxation for Empty Result Sets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Customers often enter overly constrained search strings on merchant websites, leading to no matching products being found, which can discourage purchases and result in lost sales for online merchants, as well as present irrelevant results due to broad queries generated from partial search strings.
Innovation Solution
A search engine detects when a query fails to identify a predefined threshold of items and generates altered versions of the search string by modifying it, such as removing terms, applying weights, and performing spell corrections, to submit second queries that identify relevant products, presenting these alterations along with their corresponding results to the user.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the search engine uses the exact search string entered by the user, then the search results are precisely matched to user intent, but no results are returned when the search string is overly constrained
Solution Approach 1:
The search system dynamically adjusts the search string by generating multiple altered versions with different levels of constraint relaxation. Instead of using a fixed search approach, the system creates a spectrum of queries from exact match to progressively relaxed versions, allowing the search precision to adapt based on result quality.
Solution Approach 2:
The system changes parameters of the search string by removing terms, applying different weights, and performing spell corrections to generate altered versions. This parameter modification allows the search to transition from highly constrained to less constrained states, resolving the contradiction between precision and result quantity.
2Quantity of substance
If the search engine relaxes the search constraints to return more results, then the number of results increases, but irrelevant results are included
Solution Approach 1:
The system segments the search results into multiple groups, each corresponding to a different altered version of the search string. This segmentation allows users to see results at different levels of relaxation and select the group that best matches their intent, maintaining precision while increasing overall result quantity.
Solution Approach 2:
The system performs partial relaxation of search constraints by generating altered versions that remove only some terms or apply selective weight changes. This partial action approach allows the system to increase result quantity while maintaining adequate precision by not over-relaxing the constraints.
3Adaptability or versatility
If the search engine presents multiple altered search strings and groups of results, then users can find relevant products even with overly constrained queries, but the interface complexity increases
Solution Approach 1:
The system automatically generates and presents multiple altered search strings and their corresponding result groups without requiring user intervention to reformulate queries. This self-service approach increases search adaptability while managing interface complexity by automating the query transformation process.
Solution Approach 2:
The system performs preliminary actions by automatically generating altered search strings and organizing results into groups before the user views them. This preliminary organization reduces the complexity the user faces by presenting pre-processed, structured information rather than raw unorganized results.
Data Source
AI summary
Disclosed are various embodiments for generating search results when a search string fails to return results as entered. In a representative example, a failure of a first query generated from a search string to identify at least a predefined threshold number of items from among a plurality of items in a data store is detected in a computing device. A plurality of altered versions of the search string is generated in the computing device. A submission of a plurality of second queries to the data store is implemented, where each of the second queries corresponds to one of the altered versions of the search string. A presentation is generated to be rendered of at least two separate groups of items resulting from corresponding ones of the second queries.