Personalized Search Ranking via Deep Attribute Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users interacting with e-commerce platforms through non-visual interfaces, such as voice input, face challenges in finding desired products among numerous search results, as they lack the ability or patience to review hundreds of listings presented through text-to-speech systems.
Innovation Solution
A system that generates personalized search results by analyzing user historic data and query attributes, using a computing device to receive search queries, generate relevance-based results, and extract item and query attributes, thereby prioritizing user-specific items in the search outcomes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional visual web client search results are presented through text-to-speech systems for voice input interaction, then users can access e-commerce inventory through alternative interfaces, but users cannot efficiently review hundreds of listings to identify desired products
Solution Approach 1:
The system segments the search results by extracting key attributes (price, color, style, brand) and presenting them as structured data elements that can be efficiently processed by text-to-speech systems, breaking down the overwhelming list into manageable informational units
Solution Approach 2:
The system changes the parameter representation by converting visual search results into spoken attribute-based descriptions, transforming the interaction mode from visual scanning to auditory attribute comparison, which better suits voice-based interfaces
2Measurement precision
If personalized search results are generated using deep attribute extraction and attentive user interest embeddings, then search result relevance to user preferences is improved, but system complexity increases
Solution Approach 1:
The system performs preliminary attribute extraction from search results and pre-computes user interest embeddings based on historic data before the actual search query, so that when a search is executed, the personalized ranking can be quickly applied without real-time computation overhead
Solution Approach 2:
The system creates simplified copies of user profiles and item attributes in embedding form, which capture the essential characteristics needed for personalized ranking without requiring access to the full complex data structures, enabling efficient comparison and matching
Data Source
AI summary
In some examples, a system may be configured to generate one or more query attributes for a search query received from a computing device of a user. Additionally, the system may be configured to, based at least in part on historical data of the user including data characterizing one or more items associated with the user, generate relevant item data. In various examples, the relevant item data characterizing a set of relevant items. Moreover, the system may be configured to, based on the relevant item data, the historical data of the user and the one or more query attributes, implement a set of operations that generate a set of personalized search results associated with the search query.