Iterative Image Search Algorithm Using Dynamic Tag Refinement

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current recommendation systems fail to accurately predict a user's current interest or preference based on historical data, leading to information overload and inability to identify specific cravings, as they lack the granularity to account for fleeting and specific desires.

Innovation Solution

A computer-implemented method that analyzes tags associated with a sequence of images presented to a user, allowing for iterative and dynamic adjustment of search criteria based on user inputs to guide them to their current interest, by processing tags and selecting subsequent images that match user preferences.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If recommendation systems analyze user history to make recommendations, then they can provide generalized recommendations based on user preferences, but they fail to accurately predict current specific interests or cravings

Engineering Contradiction:

Improveprecision of current interest predictionVSAvoidgranularity of current preference

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent implements a dynamic iterative search algorithm that adapts to the user's current state in real-time. Instead of relying on static historical data, the system continuously refines recommendations based on user feedback during the interaction session, allowing it to capture fleeting current interests that differ from general preferences

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system incorporates real-time user feedback mechanisms where user interactions with presented items are fed back into the algorithm to refine subsequent recommendations. This feedback loop enables the system to adjust to current specific interests dynamically, resolving the contradiction between using historical data and capturing current cravings

Inventive Principle:

Principle #23Feedback

2Adaptability or versatility

If the system presents multiple images and options to the user, then it can explore various possibilities, but it creates information overload that prevents quick identification of specific interests

Engineering Contradiction:

Improveability to explore user interestsVSAvoidtime to identify current interest

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent segments the recommendation process into iterative stages, presenting a limited number of images at each step rather than overwhelming the user with all options simultaneously. This segmentation allows the system to explore various possibilities while maintaining manageable information flow at each decision point

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system employs partial action by presenting a subset of relevant images at each iteration rather than all possible options. This approach balances exploration of user interests with efficient time consumption, as the iterative nature allows convergence on specific interests without requiring the user to evaluate every possible option

Inventive Principle:

Principle #16Partial or excessive action

3Productivity

If the system uses general user preferences for recommendations, then it can provide broad recommendations, but it lacks the granularity to account for specific current cravings

Engineering Contradiction:

Improvespeed of recommendation deliveryVSAvoidgranularity of preference matching

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system dynamically adjusts the granularity of preference matching during the interaction. It begins with broader categories and progressively refines to more specific preferences based on user feedback, enabling both quick initial recommendations and increasingly precise matching as the session progresses

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent implements preliminary action by using general preferences to generate initial recommendations quickly, then refining based on user feedback. This allows the system to deliver recommendations at speed while progressively improving granularity, rather than requiring full precision from the start

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11669557B2Iterative image search algorithm informed by continuous human-machine input feedback

Publication Date: 2023.06.06 ASK SYDNEY LLC

AI summary

System and computer-implemented method of analyzing tags associated with a sequence of images presented to a user to present a current interest of the user is disclosed. An image from among a plurality of images is presented on an electronic display. The image is associated with a set of tags. An input is received indicating a user's preference for the image. A plurality of tags is processed based on the preference and the set of tags to determine a next set of tags from the plurality of tags. A next image is determined from the plurality of images based on the next set of tags. The next image represents a physical object, different from a physical object represented by the previous image. A sequence of images is generated by repeating the above process with the next image in place of the previous image for present a user's current interest.