BlossomPsy 用户为中心的自适应与互动 MBTI 人格评测系统

Jul 16, 2026

Upload by Oliver

BlossomPsy presents a user-centric, AI-driven MBTI assessment system designed to improve public accessibility and participation while preserving predictive stability. The approach builds on the MBTI framework and combines multi-turn dialogue with photo-based questions to increase interactivity and supply additional user information. Using deep learning, multi-armed bandit methods, and control-theoretic feedback, the system adapts dynamically across user responses. Experiments with human volunteers and LLMs indicate stable predictions and higher reported user satisfaction versus MBTI-M (Chinese), with comparable consistency to the reference scale.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages23

Reading Time58 min

Introduction

Public interest and limitations of traditional assessments

Prior interactive and AI-supported assessment work

BlossomPsy: system design and two-stage prediction strategy

arXiv :2607 .06 149v2 [ cs .HC] 8 Jul 2026 BlossomPsy: A User-Centric AI System for Adaptive and Engaging MBTI Personality Assessments Bingjia Huang University of Electronic Science and Technology of China Abstract There has been growing public interest in understanding personality traits and emotional characteristics, as such knowledge helps individuals better accept themselves and manage negative emotions. While professional personality scales remain the standard tool for assessment, they are often perceived as tedious or inaccessible to the general public. AI-driven systems can make assessments more accessible, but it is difficult to balance user engagement with predictive consistency in existing works. We tackle this challenge by introducing BlossomPsy, a userfriendly AI-driven MBTI assessment system. MBTI, a widely recognized but psychometrically debated personality framework, serves as the foundation for many recent systems. BlossomPsy integrates multi-turn dialogue and photo-based questions to enhance user engagement while supporting confidence-aware predictions. By combining deep learning, multi-armed bandit algorithms, and control theory, the system dynamically adapts to users’ responses. In particular, photo-based questions are designed to increase interactivity and provide additional user information, thereby improving prediction confidence. Experiments involving both human volunteers and large language models (LLMs) provide preliminary evidence that BlossomPsy can produce stable predictions, with higher reported user satisfaction compared to MBTI-M (Chinese version), while maintaining comparable consistency with the reference scale. 1 Introduction Over the years, public awareness of mental health has steadily increased [1] . By knowing themselves better, people are more able to recognize and handle emotional challenges. Yet, most psychological assessment tools still use formats that have remained largely unchanged for decades. Typically, individuals need to complete a set of choice questions, after which the system automatically calculates their scores [2–4] . By comparing their scores with standardized manuals, the system can provide a structured result to people. However, this kind of assessment often suffers from limited user engagement, repetitive formats, and difficulty adapting to individual differences. Many users consider such assessments tedious, leading to low participation especially among young people [5] . To make the assessment engaging, psychologists have explored various interactive or visual-based methods for assessments over the decades. Buck [6] put forward the House-Tree-Person Test (HTP) . In the test, individuals draw a house, a tree and a person; the drawings are then analyzed to gain insights into their personality, emotions and inner experiences. Greenwald et al. [7] introduced Implicit Association Test (IAT), in which the strength of automatic associations between concepts is measured to assess potential bias or attitude. Further, game-based assessments like [8] are used to assess children’s cognitive control. Recent advances in artificial intelligence and human-computer interaction have created new opportunities to reimagine psychological testing. Some studies have used NLP techniques to support psychology evaluation, like empathy detection [9], mental illness and agony evaluation [10] and text-based personality assessment [11, 12] . Teferra et al. [13], Zhang et al. [14] provide a general review of NLP-based psychology evaluation. Although these studies have opened up a new avenue Preprint. for exploration, they often lack practical applicability [15] . LLMs may further improve applicability. Some researchers have aimed to improve the capabilities of large language models (LLMs) so that they can provide psychological assessment traditionally handled by psychologists [16, 17] . Related research includes psychological disorder detection [18, 19], mental health counsel

FAQs

Why are traditional MBTI-style assessments considered difficult for many users?

They often rely on repetitive choice-question formats that feel tedious and are less adaptable to individual differences, which can reduce participation, especially among younger users.

How does BlossomPsy improve user engagement during MBTI assessment?

It uses multi-turn dialogue plus photo-based questions to make the interaction more engaging and to provide additional user information for better prediction confidence.

How does BlossomPsy maintain prediction reliability while adapting to users?

It combines confidence-aware predictions with a fallback mechanism that presents photo-based single-choice questions when confidence is insufficient, while the conversation strategy is guided by a modified multi-armed bandit algorithm and PID feedback control.

Jul 16, 2026

Upload by Oliver