Client-Side Search Result Personalization via Local Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search technologies often fail to personalize search results effectively without compromising user privacy, as they require transmitting sensitive information to servers, and may not prioritize relevance based on recent user behavior and context.
Innovation Solution
Implementing client-side personalization on user devices, which scores and orders search results based on relevance to recent user behavior and context, while maintaining privacy by processing data locally and presenting information about the source of each result.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If server-side personalization is implemented, then search result personalization is improved, but user privacy is compromised due to transmission of sensitive information
Solution Approach 1:
Instead of sending user data to the server for personalization, the patent inverts the approach by sending search results from the server to the client device, where personalization is performed locally using user profile data stored on the client. This reversal of the personalization location eliminates the need to transmit sensitive user information while still achieving personalized search results.
Solution Approach 2:
The patent introduces an intermediary mechanism where user profile information is stored and processed locally on the client device rather than being transmitted to the server. This local storage acts as an intermediary that enables personalization without requiring direct transmission of sensitive user data between client and server, thus protecting user privacy while maintaining personalization capability.
2Adaptability or versatility
If search results are personalized by the server, then personalization capability is improved, but network transmission of sensitive data increases
Solution Approach 1:
The patent applies the inversion principle by reversing the location where personalization occurs. Instead of the server personalizing results using transmitted user data, the server sends raw search results and the client device performs personalization locally using stored user profiles. This eliminates the need to transmit sensitive user information over the network while maintaining full personalization capability.
Solution Approach 2:
The patent extracts the personalization function from the server side and relocates it to the client side. By taking out the personalization processing from the server, the system eliminates the requirement to transmit sensitive user data over the network, as personalization is performed using locally stored user profile information on the client device.
3Object-affected harmful factors
If client-side personalization is implemented, then user privacy is protected, but processing complexity on user device increases
Solution Approach 1:
The patent applies preliminary action by pre-storing user profile information and search result data locally on the client device before personalization is needed. User profiles, search history, and search results are cached in advance, so when personalization is required, the client device can perform scoring and ranking operations using pre-available data without requiring complex real-time processing or additional data retrieval.
4Measurement precision
If search results are scored and reordered based on user behavior, then search result relevance is improved, but computation time on client device increases
Solution Approach 1:
The patent applies partial action by implementing a two-stage scoring system. First, the server performs initial scoring and filtering of search results based on user profiles before transmission. Then, the client device performs additional local scoring and reordering. This partial division of computation work between server and client achieves high result relevance while minimizing the computational burden and time required on the client device, as the most time-consuming scoring operations are already completed by the server.
Data Source
AI summary
In some implementations, a user device (e.g., a computing device) can perform client-side personalization of search results. For example, a computing device can obtain search results matching user specified search parameters from a server device and/or from various services on the user device. The user device can score the search results based on various search result item attributes. After scoring, the user device can promote or demote search results items based on whether the search results item is relevant to recent user behavior. The promotion and/or demotion of search results items can cause search results items scores to be adjusted to generate a personalized score for each search result. The search results can then be ordered and/or presented based on the personalized score for each search results item. When presenting search results items, the user device can present information indicative of the source of the search results items.