Intelligent Navigation System Using Feedback-Validated Editor Recommendations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current e-commerce navigation systems face challenges in providing relevant and rich category recommendations due to noise interference, misplacement of categories, and lack of user feedback, leading to inaccurate and incomplete search results, which complicates user shopping experiences.
Innovation Solution
A method and apparatus for intelligent navigation that utilizes a navigation dictionary incorporating editor recommendations based on user behavior data, including search logs, clicking logs, and purchase logs, to enhance recommendation accuracy and relevance by generating index keywords, categories, and content, and integrates user feedback for validation and modification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If bottom-up recommendation algorithm is used to generate category recommendations, then the system can automatically provide navigation suggestions, but noise interference and misplacement of categories reduce recommendation accuracy
Solution Approach 1:
The patent introduces an intermediary validation mechanism where user feedback serves as a mediator between the automated recommendation system and the final recommendation output. The system collects user feedback on recommended categories and uses this feedback to validate and adjust the automated recommendations, thereby reducing noise interference and misplacement issues while preserving automation.
Solution Approach 2:
The patent implements a feedback loop where user interactions with recommended categories are collected and used to improve future recommendations. By incorporating user feedback into the recommendation algorithm, the system continuously refines its accuracy and reduces the impact of noise and misplacement in automated category suggestions.
2Device complexity
If traditional hierarchical category structure is used for navigation, then the system maintains simple data organization, but users need to click multiple times to reach specific categories increasing search time
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing optimal navigation paths from various starting categories to target categories. When a user initiates a search, the system retrieves these pre-computed paths and presents them as recommended categories, allowing users to jump directly to relevant categories without sequential clicking through the hierarchical structure.
Solution Approach 2:
The patent adds a new dimension to the traditional hierarchical category structure by introducing recommended category paths that span across multiple hierarchical levels simultaneously. Instead of requiring users to traverse the hierarchy level by level, the system provides direct cross-level navigation suggestions, effectively adding a shortcut dimension to the category navigation space.
3Measurement precision
If editor manual editing is used to add artificial recommendations, then recommendation relevance can be improved, but the process lacks user feedback and traceability
Solution Approach 1:
The patent implements feedback mechanisms that track user interactions with editor-recommended categories. User feedback data is collected and stored, creating a traceable record of how manual recommendations perform in practice. This feedback loop allows editors to refine their recommendations based on actual user behavior, improving relevance while maintaining traceability.
Solution Approach 2:
The patent introduces user feedback data as an intermediary between manual editing and recommendation deployment. Rather than directly applying editor recommendations without validation, the system uses user feedback as a mediator to validate and adjust manual recommendations, ensuring both relevance and traceability in the recommendation process.
4Measurement precision
If dynamic category presentation based on user behavior scoring is implemented, then category relevance is improved, but the system becomes complex and requires processing large amounts of user data
Solution Approach 1:
The patent applies local quality by implementing user behavior scoring selectively for specific categories and user contexts rather than uniformly across all categories. The system identifies key categories where dynamic presentation provides the most value and applies complex scoring mechanisms only there, while maintaining simpler presentation for other categories, thereby reducing overall system complexity.
Data Source
AI summary
The present disclosure describes a method, an apparatus and a system of intelligent navigation. In one embodiment, a method includes: receiving a user inquiry from a client terminal; searching a navigation dictionary based on the user inquiry to obtain a recommendation result corresponding to the user inquiry, the navigation dictionary including an editor recommendation based on user behavior information; and sending the recommendation result to the client terminal. The present disclosure can enhance the accuracy, relevancy, richness and intelligence of the intelligent navigation, and reduce user search time as well as the search loading on the server.