Session Context Embeddings for Accurate Search Result Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search query technologies often fail to capture the user's intent accurately due to single query-item pair relevance, leading to irrelevant search results.
Innovation Solution
A system that tracks electronic interactions during a user session to generate session context embeddings, which are used to query a vectorized item database and improve the relevance of search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If single query-item pair relevance training is used, then the search system is simple and fast, but the user intent is not captured accurately leading to irrelevant search results
Solution Approach 1:
The system performs preliminary actions by tracking user interactions and generating session context embeddings before the actual search query is processed. User behavior data is collected and encoded into embeddings that capture intent context in advance, allowing the search system to leverage this pre-computed context for more accurate results without adding complexity to the core search indexing mechanism.
Solution Approach 2:
Session context embeddings serve as an intermediary between raw user interactions and search results. Instead of directly analyzing complex user behavior patterns during search, the system uses these embeddings as a condensed representation of user intent that bridges the gap between simple query processing and sophisticated intent understanding, enabling accurate results while maintaining system simplicity.
2Measurement precision
If session context embedding is generated from tracked interactions, then search result relevance is improved, but processing time and computational resources increase
Solution Approach 1:
Session context embeddings are generated in advance as users interact with the system, before the actual search query is submitted. This preliminary computation of intent representations from user behavior data allows the search system to quickly retrieve and rank results using pre-computed embeddings, significantly reducing real-time processing requirements while maintaining high relevance accuracy.
3Measurement precision
If multiple user interactions are tracked to understand intent, then search accuracy improves, but the system complexity and data processing requirements increase
Solution Approach 1:
The system extracts and isolates the essential intent information from complex user interaction sequences by generating session context embeddings. This extraction process separates the meaningful intent signals from the noise of redundant interaction data, allowing the system to focus only on processing the condensed embedding representations rather than analyzing every individual user action, thereby reducing overall system complexity.
Solution Approach 2:
Multiple user interaction signals are merged and combined into a single session context embedding that represents overall user intent. By consolidating diverse interaction data (clicks, views, dwell time, etc.) into one unified embedding representation, the system simplifies processing while maintaining comprehensive intent understanding, avoiding the need to process each interaction type separately.
Data Source
AI summary
A method can include: storing, on an electronic platform, a vectorized item database corresponding to a plurality of items; tracking electronic interactions between the electronic platform and a user computer during a user session; generating, using an embedding model, a session context embedding based, at least in part, on the electronic interactions tracked during the user session; in response to receiving the search query from the user computer, querying the vectorized item database to identify search results corresponding to one or more of the plurality of items, at least in part, using the session context embedding; and transmitting, by the electronic platform, the search results to the user computer. Other embodiments are disclosed.