Dynamic Query Interpretation for Search Refinement Recommendations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users often face challenges in finding specific products on e-commerce marketplaces due to inadequate search terms and difficulty in using filter features, leading to multiple searches and increased computational load on the marketplace infrastructure.
Innovation Solution
A search recommendations system that utilizes machine learning models to generate refinement and search recommendations based on initial search terms, incorporating factors like semantic similarity, user interaction data, and merchant locations to assist users in narrowing or changing their search effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If users perform multiple searches to find specific products, then search completeness is improved, but computational load on marketplace infrastructure increases
Solution Approach 1:
The system performs preliminary action by generating and pre-ranking multiple potential search recommendations before the user completes their search query. The machine learning model analyzes the initial search terms and proactively generates refined search queries that are likely to lead to desired products, reducing the need for users to perform multiple separate searches and thereby decreasing overall computational load while maintaining search completeness.
Solution Approach 2:
The patent introduces an intermediary mechanism in the form of a machine learning recommendation system that acts as a mediator between the user's initial search terms and the final product results. This intermediary generates and ranks multiple potential search refinements, effectively bridging the gap between simple search queries and comprehensive product discovery, reducing the iterative search process and associated computational overhead.
2Measurement precision
If users perform multiple searches to find specific products, then search accuracy is improved, but time required for shopping increases
Solution Approach 1:
The system applies preliminary action by pre-generating and ranking multiple accurate search refinements based on the user's initial search terms before the user needs to execute the search. The machine learning model analyzes search patterns and product data to proactively generate accurate refinements that are likely to yield desired results, reducing the time users would otherwise spend iteratively searching and refining their queries.
Solution Approach 2:
The patent implements feedback mechanisms where the system continuously learns from user interactions with search results and adjusts its recommendation algorithm accordingly. This feedback loop enables the system to improve search accuracy over time by adapting to user preferences and search patterns, thereby reducing the time required for shopping as the system becomes more adept at predicting desired product results.
3Measurement precision
If filter features are made more comprehensive, then product selection precision is improved, but ease of operation decreases
Solution Approach 1:
The patent introduces an intermediary recommendation system that mediates between comprehensive filter options and user ease of operation. Instead of requiring users to manually navigate through numerous filter categories and settings, the machine learning model analyzes the user's search terms and product interests to automatically generate and present the most relevant filter combinations, maintaining precise product selection while simplifying the user interface and operation.
Solution Approach 2:
The system applies self-service by enabling the filter recommendation mechanism to automatically adapt to user preferences and search contexts without requiring manual configuration. The machine learning model self-adjusts filter recommendations based on real-time user behavior and search patterns, providing precise product selection automatically rather than requiring users to manually configure complex filter settings.
Data Source
AI summary
Examples provide improved methods for generating search recommendations in response to a user-initiated search. Examples include receiving a search request including search terms; identifying one or more product categories as output from a machine learning classification model; identifying products that are assigned to those product categories, including product titles short descriptions in a natural language; applying the product titles and short descriptions as input to a second machine learning model that is configured to generate recommended searches; scoring each recommended search of the plurality of recommended searches; selecting one or more recommended searches of the plurality of recommended searches based on the scoring; and causing the one or more recommended searches to be displayed as user-interactable components on a graphical user interface, each user-interactable component being configured to execute a second search request upon user interaction with the user-interactable component.