User Feedback Trust Scoring System Based on Cross-Shaped UI

The cognitive-based verification module with a cross-shaped UI and OX-based verification addresses positional biases and unconscious errors, enhancing user feedback reliability and AI training data quality through a cyclic learning feedback loop.

KR1020260161682APending Publication Date: 2026-11-10장서은
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
KR1020250057863
Authority / Receiving Office
KR · KR
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-05-01
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing AI user feedback systems suffer from positional biases, unconscious selection errors, lack of reliability evaluation, and inadequate interface design for inducing user participation and feedback incentives.

Method used

A cognitive-based verification module with a cross-shaped UI array and OX-based structural difference verification, along with confidence-based feedback weighting, to ensure objective responses and improve AI training data quality.

Benefits of technology

Enhances user feedback reliability and AI training data quality by eliminating positional biases and unconscious errors, while quantifying feedback quality and improving AI through a cyclic learning feedback loop.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure PAT00001_ABST
    Figure PAT00001_ABST
Patent Text Reader

Abstract

The present invention relates to a structure for improving the quality of user feedback in an artificial intelligence (AI) system. In particular, it implements a cyclic learning system that prevents positional bias through a cross-shaped (four-part) UI arrangement, verifies user understanding through a cognitive-based OX verification procedure, and assigns feedback weights based on reliability scores. Through this, user experience and AI learning accuracy can be improved simultaneously, and it is applicable to various industries such as education, finance, and counseling.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a user feedback processing technology for an artificial intelligence (AI) system, and more specifically, to a user interface (UI) design, a cognitive-based verification module, and a reliability weighting application method for solving problems such as positional bias, unconscious selection errors, and reduced feedback reliability. Background Technology

[0003] Existing AI user feedback systems generally use a selection structure where A / B options are placed vertically or horizontally, or utilize collection methods based on external survey links, but they have the following limitations.

[0004] First, unconscious positional biases based on up / down or left / right arrangement occur repeatedly, distorting the response.

[0005] Second, due to the lack of a verification procedure regarding actual awareness, it is difficult to quantitatively evaluate the reliability of the selected feedback.

[0006] Third, there are limitations in inducing repeated participation due to the lack of interface design and user reward systems capable of stimulating incentives for writing feedback. The problem to be solved

[0008] The present invention aims to solve the following problems.

[0009] (1) Ensure objectivity of the response by eliminating recurring positional bias and primacy effects

[0010] (2) Includes a cognitive verification mechanism to verify whether the judgment is based on the user's actual understanding.

[0011] (3) Implement an algorithm structure that can quantitatively evaluate feedback quality and incorporate reliability-based weighting.

[0012] (4) The entire structure is configured as a single cyclic learning feedback loop to autonomously improve the internal quality of AI over the long term. means of solving the problem

[0014] The present invention solves the above problem through the following technical components.

[0015] First, cognitive equivalence among the options is ensured by eliminating visual weight and alignment bias through the even placement of response items in a four-part cross shape on the screen. This configuration is based on a visual alignment structure similar to a test paper, thereby minimizing primacy effects and positional bias.

[0016] Second, an OX-based structural difference verification module is introduced to confirm whether the user's choice is a judgment based on cognition. Immediately after selecting a response, users can participate in a question asking about the structural differences between the selected response and other responses, and participation is provided optionally.

[0017] Third, a reliability score is assigned to the user based on the verification results, and the quality of the training data is enhanced by reflecting weights in the user's feedback. Users who secure a reliability score above a certain standard are classified as high-reliability users and are designed to participate in advanced tests and additional features.

[0018] This series of configurations forms the entire system into a structure that is cyclic and learns autonomously, and can continuously improve the internal quality of the AI. Effects of the invention

[0020] Quantification of user feedback reliability and automatic classification possible

[0021] Systematic elimination of positional bias and unconscious choice errors

[0022] Enhance user UX and immersion by providing an iterative learning structure through true / false verification

[0023] Implementation of an autonomous learning structure that continuously improves the quality of AI training data

[0024] Securing a foundation for generating advanced secondary information with universality applicable across industries

[0025] Particularly suitable for areas where AI is utilized to assist human decision-making, such as education, finance, and counseling. Brief explanation of the drawing

[0027] The attached diagram below is a schematic representation visualizing the flow experienced by a user in the proposed system. Each step includes key elements for improving AI quality and is composed of an interaction-based structure. Specific details for implementing the invention

[0028] The present invention consists of the following three-step mechanism.

[0029] (1) Cross-shaped UI array

[0030] Response items are evenly distributed within a four-part cross-shaped structure on the screen. This forms a visual structure similar to an exam paper, ensuring that all options start on the same level to minimize psychological primacy effects and variations in visual weight. Button animations and alignment elements are also aligned to realize a user interface design that blocks positional bias.

[0031] (2) OX-based structural difference verification

[0032] Immediately after a user selects a response, a true / false question is presented asking about the structural differences between the selected answer and other responses. This method evaluates whether the choice is based on actual understanding rather than mere preference or habit. (e.g., "Responses 1 and 3 utilize data tables differently.") This approach is based on the psychological premise that selection and cognition are distinct processes, and the quiz-style true / false verification contributes to learning reinforcement through cognitive rewards for the user. This verification operates on a voluntary participation basis, encouraged by phrases such as "Could you help a little more?", and provides a reward message upon participation. (e.g., "Your learning has been reinforced by 3 times, and your answer contributed 5 times to AI improvement.")

[0033] (3) Apply confidence-based feedback weights

[0034] AI quality is improved by giving higher weight to feedback from users who recognize the difference through OX verification and respond accurately. Users who achieve a certain accuracy rate or higher can be classified into the 'Trusted Feedback Group' to encourage their participation in advanced tests.

Claims

Claim 1 A user feedback interface system for eliminating positional bias and ensuring objectivity of responses when inputting user feedback, comprising: (a) a user interface (UI) module that prevents visual primacy effects and positional bias occurring in vertical or horizontal arrangements by arranging multiple response items evenly in a cross-shaped (four-part) structure on the screen, wherein the UI module unifies the visual weight, alignment direction, and interaction elements (button animations, etc.) of each response item, thereby inducing the user to select all response items under cognitively equal conditions. Claim 2 A reliability-based user feedback evaluation system according to claim 1, comprising: (a) presenting a verification question in the form of an OX to check whether the user is aware of differences in structure, content, etc. between the selected response and other responses immediately after the user selects a response item; (b) evaluating the user's level of awareness based on whether the response to the verification question is correct or incorrect; and (c) assigning different reliability-based weights to the response feedback selected by the user according to the evaluation result.