Language-Independent Recommendation System for Contextual Search

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional recommendation systems fail to develop contextually relevant recommendations and are limited by language and meaning constraints, as well as being confined to specific sites where user information is collected, leading to suboptimal search results and product suggestions.

Innovation Solution

A recommendation system that customizes search results and product lists based on analyzing previous actions and behaviors of users, using language-independent modeling to deliver contextually relevant content by scoring user interactions and associating them with content items, and matching current users with similar behaviors to provide personalized recommendations across various environments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If conventional recommendation systems use cookies to track user online activity, then user behavior can be monitored, but the system fails to develop contextually relevant recommendations and is limited to specific sites

Engineering Contradiction:

Improveuser behavior informationVSAvoidcontextual relevance of recommendations

Core Design Contradiction:

Loss of informationVSAdaptability or versatility

Solution Approach 1:

The patent segments user behavior tracking into multiple independent data collection mechanisms including cookies, device identifiers, and cross-site tracking technologies. Each mechanism captures specific aspects of user behavior that can be independently processed and combined to create comprehensive user profiles, thereby overcoming the limitations of single-site tracking while maintaining contextual relevance

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements a universal recommendation system that functions across multiple websites and platforms using the same core technology. The system uses multi-functional tracking mechanisms that work across different sites and devices, enabling consistent contextual recommendation delivery regardless of where the user interacts, thus resolving the site-specific limitation

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Ease of operation

If conventional systems rely on language-specific search and tagged knowledge bases, then search functionality is provided, but customized content is missed due to language and meaning constraints

Engineering Contradiction:

Improvesearch functionalityVSAvoidcontextually relevant content

Core Design Contradiction:

Ease of operationVSLoss of information

Solution Approach 1:

The patent replaces traditional language-based mechanical search systems with a behavior-based recommendation system. Instead of relying on keyword matching and tagged knowledge bases, the system substitutes these with algorithms that analyze user behavior patterns, clicks, and interactions to infer intent and deliver relevant content, thereby overcoming language and meaning constraints while maintaining search functionality

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Loss of information

If the system captures and aggregates extensive user behavior data across multiple sites, then more comprehensive user models can be developed, but system complexity increases

Engineering Contradiction:

Improveuser behavior data completenessVSAvoidsystem architecture

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The patent introduces intermediary components including centralized data aggregation servers, behavior analysis engines, and profile management systems that mediate between multiple data collection points and the recommendation delivery mechanism. These intermediaries standardize and process raw behavior data from various sources, reducing overall system complexity while maintaining comprehensive data capture

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements a nested system architecture where user profiles contain nested behavior patterns, which contain nested contextual information. This hierarchical nesting organizes extensive user data in a structured manner, making complex information manageable and accessible without increasing operational complexity at each level

Inventive Principle:

Principle #7Nested doll (Nesting)

Data Source

PatentUS9881332B2Systems and methods for customizing search results and recommendations

Publication Date: 2018.01.30 LOGOMIX INC(US)

AI summary

A recommendation system can be configured to customize search results and/or recommendations of content (e.g., customized products, products, advertising, layouts, etc.) using online and/or offline activity captured on a user population. The system can be configured to customize the content returned to users to achieve specific behaviors and/or influence the current user's behavior, responsive to modeling previous users. For example, the system can capture and aggregate user behavior/activity and score content based on the actions taken with respect to the content. In some examples, the scoring can be filtered or augmented by matching current user characteristics to characteristics of the previous users. The scoring can be generated independent of the language in which the activity/user behavior occurred. According to one embodiment, the system is configured to generate language independent models and utilize the language independent modeling to deliver customized content (e.g., recommendations and/or search results) without conventional constraints.