Skip to main content

XJTLU CPT208  ·  Group B1-6

X-Thread

"Weaving Thoughts into a Net, Thread by Thread"

Light up to enter

X-Thread

Live Demo
CPT208 Human-Centric Computing | Group B1-6 | Xi'an Jiaotong-Liverpool University

Project Theme

X-Thread is a platform dedicated to fostering more meaningful, in-depth, and equitable group discussions among XJTLU students in academic and language classrooms.

We integrate GenAI, collaborative workspaces, and structured workflows to break time/space barriers and transform fragmented discussions into structured, recordable, and synchronized collaboration processes.

Introduction

Why This Track?

We chose the B1 XJTLU Group Discussions track because, as XJTLU students ourselves, we have lived through the very frustrations this project addresses. We believe the intersection of playful interaction design and collaborative learning presents a unique opportunity to do more than build another meeting tool — it lets us reimagine the entire discussion experience as something students genuinely want to participate in.

Existing Problems

Time Conflicts

Hard to schedule synchronous collaboration; on-campus study spaces often fully occupied

😰

Awkward Icebreaking

Unfamiliar team members → awkward atmosphere → difficult to start topics → low efficiency

Lack of Academic Support

No real-time academic support; complex questions get stuck; can only ask teachers/TAs after class

📝

Disorganized Records

No structured records; key ideas easily lost; messy notes; hard to find discussion logic during review

Opportunities

GenAI + Collaboration Tools can solve these problems:

Project Aims

Break Geographical & Temporal Barriers

Empower XJTLU students with a flexible, hybrid collaboration platform that eliminates scheduling and location barriers, enabling them to join group discussions anytime, anywhere.

Deliver Real-Time AI-Powered Support

Provide instant, context-aware academic support through integrated GenAI, helping students resolve questions and clarify concepts in real time to keep discussions productive.

Preserve & Organize Key Insights

Enhance idea retention and post-meeting accountability with collaborative mind mapping and shared whiteboarding tools, ensuring key insights are captured, organized, and easily revisited.

Streamline & Boost Participation

Eliminate awkward silences and low engagement in randomly-assigned teams with guided onboarding and structured workflows, reducing social friction and helping groups dive into productive discussions faster.

Literature Review & Product Analysis

Key Research Papers

Online Group Projects in Higher Education: Persistent Challenges and Implications for Practice

2023

H. Donelan, K. Kear

A comprehensive survey of real student experiences with online group projects, identifying persistent challenges across multiple institutions and proposing evidence-based recommendations grounded in practitioner experience.

Advantages
  • Provides authentic student perspectives from diverse institutional contexts
  • Identifies recurring pain points (coordination, communication, free-riding) with clear evidence
  • Offers practical, implementable recommendations for educators and platform designers
Shortcomings
  • Focuses primarily on asynchronous formats; limited discussion of synchronous real-time collaboration
  • Does not explore technological interventions (AI, gamification) as potential solutions
  • Data predates widespread adoption of GenAI tools in higher education

LLM-based Collaborative Agents with Pedagogy-guided Interaction Modeling for Timely Instructive Feedback

2025

Y. Qihao et al.

Pioneers the integration of LLM-based agents with pedagogical interaction modeling to provide timely, instructive feedback in task-oriented group discussions, demonstrating measurable improvements in discussion quality.

Advantages
  • Novel combination of LLM agents with established pedagogical frameworks
  • Provides a concrete technical architecture for real-time AI-driven feedback generation
  • Demonstrates measurable improvements in task-oriented discussion outcomes
Shortcomings
  • Limited to task-oriented discussions; may not generalize to open-ended academic discourse
  • Requires significant computational infrastructure not available to all institutions
  • Does not fully address potential risks of student over-reliance on AI feedback

Collaborative Concept Maps in Higher Education: Pedagogical Contributions, Cognitive Challenges, and Optimization Strategies

2025

Lotfi. F. Z. et al.

A comprehensive review of collaborative concept mapping in higher education, identifying specific cognitive challenges students face and proposing actionable optimization strategies for interactive visual learning tools.

Advantages
  • Thorough review linking pedagogical theory to practical concept mapping implementation
  • Identifies specific cognitive challenges (cognitive load, coordination overhead) with clarity
  • Proposes concrete optimization strategies transferable to digital collaboration platforms
Shortcomings
  • Limited empirical validation of the proposed optimization strategies
  • Focuses on concept maps as a single modality; does not explore integration with broader tools
  • Does not address accessibility considerations for students with disabilities

Usability and User Satisfaction Rate Evaluation on E-Learning Application Using Nielsen Usability Method

2021

A. Sibarani

Demonstrates the practical application of Nielsen's heuristic evaluation framework in e-learning contexts, providing clear metrics (usability score, satisfaction rate) and a replicable evaluation methodology.

Advantages
  • Applies a well-established, industry-standard evaluation framework (Nielsen's heuristics)
  • Provides clear, quantifiable metrics (usability score, satisfaction rate) suitable for benchmarking
  • Offers a replicable methodology directly applicable to e-learning platform evaluation
Shortcomings
  • Study sample limited to a single institution; findings may not generalize across contexts
  • Evaluates existing platforms rather than informing forward-looking design decisions
  • Does not incorporate complementary evaluation methods (cognitive walkthrough, A/B testing)

Competitive Product Analysis

Discord
Pros: Real-time communication, voice channels, rich media support
Cons: Lacks educational structure, limited discussion tracking, no analytics
Slack
Pros: Threaded discussions, integrations, file sharing
Cons: Designed for business, steep learning curve, expensive for education
Microsoft Teams
Pros: Integrated with Office 365, video conferencing, educational features
Cons: Complex interface, overwhelming for students, limited gamification
Canvas Discussions
Pros: LMS integration, grading tools, structured forums
Cons: Outdated UI, low engagement, no real-time features

User Journey Map

1

Discovery

Student learns about the platform through class introduction or peer recommendation

🔍
2

Onboarding

Creates account, sets up profile, joins relevant discussion groups

🚀
3

Exploration

Browses discussion topics, reads existing conversations, understands the platform

📖
4

Participation

Posts comments, responds to peers, engages in meaningful discussions

💬
5

Growth

Earns badges, builds reputation, sees personal progress and learning outcomes

🌟

Core Features

Requirements Discovery

User Interview Key Findings

1. Current Group Collaboration Workflow

1

Scheduling: Coordinate meeting times via group chat, then book offline study rooms for discussions.

2

Hybrid fallback: If in-person meetings are impossible, they use WeChat voice messages and manual notes.

3

Pre-work & problem-solving: Members study materials individually first; if stuck, they wait to ask teachers or seniors after class.

2. Key Pain Points & Inefficiencies

1

Coordination barriers: Severe schedule conflicts and fully booked classrooms often force late-night meetings.

2

Awkward group dynamics: Unfamiliar teammates create awkward silences and low-efficiency discussions.

3

Disorganized communication: Chaotic chat logs and scattered notes make key points hard to find or review.

4

Lack of real-time support: No immediate academic help leads to repeated roadblocks and wasted time.

5

Poor traceability: Manually compiled notes lack clear discussion logic, making follow-up work difficult.

3. Implication for X‑Thread

1

A structured platform to reduce scheduling and coordination friction

2

Tools to lower awkwardness and improve discussion flow

3

Built-in real-time academic support

4

Structured, auto-synced note-taking for clear progress tracking

Target Users

CR

Cristina

20, Senior 3, Financial Mathematics, XJTLU

Efficiency-Focused

A busy student balancing 3–4 weekly group projects, portfolio work, and study-abroad applications. She is familiar with hybrid learning but frustrated by unstructured, inefficient collaboration.

Key Frustrations
  • Dislikes awkward icebreakers and small talk
  • Struggles with off-topic discussions and dead air
  • Feels lost without a clear meeting structure
  • Frustrated by messy chat logs and scattered notes
HX

Hanyi.Xu

20, Senior 3, Electrical Engineering, XJTLU

Connection-Focused

A social, outgoing student who thrives in collaborative settings. He loves connecting with new teammates and often acts as the "glue" of the group, keeping conversations inclusive and lively.

Key Frustrations
  • Dislikes overly rigid meetings that kill natural conversation
  • Hates awkward silences and one-sided discussions
  • Frustrated when quiet members hold back ideas
  • Disorganized chat logs lose collaborative energy

Core User Requirements (6 Must-Haves)

1. Structured Onboarding & Icebreaking Support

Offer dedicated team matching and guided icebreaker activities to help users quickly adapt to unfamiliar teammates and reduce awkwardness.

2. Live AI-Powered Problem Solving

Provide real-time, in-meeting AI assistance to answer study-related questions and resolve confusion on the spot, preventing wasted time due to roadblocks.

3. Collaborative Mind Mapping & Idea Organization

Support live co-editing of mind maps during discussions to capture key insights in real time, with automatic post-meeting summaries for easy review.

4. Quick & Flexible Idea Logging

Include a shared note-taking board to capture spontaneous thoughts, ensuring no ideas are lost during the discussion.

5. Seamless Document Sharing & Analysis

Enable in-meeting file sharing and collaborative annotation for efficient document review and discussion, supporting post-meeting summary work.

6. Full-Featured Video Call Environment

Offer a stable, full-featured video call experience to support smooth communication for hybrid team collaboration.

Playful Design Requirements (3 Must-Haves)

A system is not "playful" simply because it has bright colors or animations. True playfulness emerges from interaction mechanics that surprise, engage, and lower social barriers. The following three core features define the playful soul of X-Thread — each inspired by game design patterns that make participation irresistible rather than obligatory.

1

Team Lobby — Game-Like Team Matching

Inspired by multiplayer game lobbies

"Transforms 'finding study partners' into an intuitive, low-friction experience — lowering the psychological barrier of initiating discussions and making random team formation both efficient and genuinely enjoyable."
Core Mechanic Gamification Team Formation Low Friction
2

Forced Sequential Speaking — "Emergency Meeting" Mode

Inspired by social deduction games like Goose Goose Duck

"Each person is allotted ~60 seconds to present their views, with the system strictly prohibiting interruptions or cross-talk — guaranteeing every member an equal opportunity to contribute."
Core Mechanic Turn-Taking Equal Voice Anti-Domination
3

In-Room AI Assistant — Real-Time Intelligence

Inspired by AI companions in modern productivity tools

"Acting as a 'silent teaching assistant,' the AI enriches discussion depth without overstepping into human interaction — augmenting intellectual quality while preserving natural social dynamics."
Core Mechanic AI-Powered Real-Time Support Auto-Summary

Field Research & Documentation

Crazy Eights Sketching

Design Iterations & Comparisons

Version 1 - Initial Concept

  • Traditional forum layout
  • Linear discussion structure
  • Basic text-only interface
  • Limited visual feedback
VS

Version 2 - Enhanced Design

  • Card-based modern UI
  • Threaded discussions with branching
  • Rich media and emoji support
  • Gamification elements

System Architecture

Presentation Layer — React 18 + Vite 5
Core React 18 TypeScript Vite 5 Tailwind CSS
State Zustand React Router v6
UI/UX Framer Motion Lucide Icons Lottie
Plugins @xyflow/react jsPDF Socket.IO Client
REST · WebSocket · WebRTC
Business Logic Layer — NestJS + Fastify
Framework NestJS Fastify TypeScript
Modules Auth Rooms Chat AI MindMap Whiteboard Files
Realtime Socket.IO WebRTC Signalling Room Gateway
Security JWT Passport Guards
Queues BullMQ Redis
Prisma ORM
Data & Storage Layer — Docker Compose
PostgreSQL 16

Primary database — users, rooms, messages, whiteboard state, mind‑map nodes

Redis 7

Session cache, real‑time pub/sub, BullMQ job queue backing store

MinIO

S3‑compatible object storage for shared files, whiteboard snapshots, exports

AI Providers
AI Service Layer — Multi‑Provider Compatible
Providers DeepSeek Kimi (Moonshot) Qwen (Tongyi) GLM (Zhipu) ModelScope
Capabilities Room Q&A Summary Generation Mind‑Map Expand Whiteboard Digest Companion Profiles

Team Responsibilities

Yulin Xia

Project Architect & Technical Lead

Tech Lead Architect

Responsible for the overall framework construction and technical architecture design of the X-Thread project, built the core service system and standardized development structure, and ensured the stable operation and logical integrity of the entire collaboration platform.

Yihan Zhang

UI/UX Prototyping Designer

UI/UX Designer

Completed the high-fidelity web prototype design of the platform, participated in the overall design and production of the project poster, focused on user interaction logic and interface usability optimization, and improved the product experience.

Yifei Li

Visual Designer & Prototype Developer

Visual Developer

Collaborated in the web interactive prototype design, independently completed the project poster design and finalization, integrated product functions with visual presentation, and guaranteed the visual unity and professionalism of the project.

Ziqian Wang

Functional Developer & Portfolio Specialist

Dev Portfolio

Developed and implemented partial core functions of the project, continuously promoted project iteration and optimization, and was responsible for the production and sorting of the project portfolio, ensuring the systematic presentation of project achievements.

Evaluation and Impact

Evaluation Methods (3 Types)

📊 SUS System Usability Scale

15 XJTLU students completed 10-question survey

82/100

Users highly recognize usability and learnability

🎯 Nielsen Heuristic Evaluation

  • Visibility of system status: File transfer, meeting, collaboration status clearly displayed
  • Recognition rather than recall: Core features directly visible, unified icons
  • Consistency and standards: Platform-wide style/interaction unified, meets usage habits

🔄 A/B Test (AI Smart Pet Icebreaking)

Prototype A

Manual trigger (@pet name) → User controllable pace, no disruption

Prototype B

Auto trigger (timed/silence popup) → High interaction but easy to interrupt

Through the simulation meeting test, the performance of the two prototypes, A and B, in terms of meeting efficiency, navigation efficiency, ease of learning, error rate, etc. was compared: Prototype B has stronger interactivity, but it is prone to interrupting discussions and distracting users' attention; Prototype A, on the other hand, enables users to control the pace independently, avoiding interference with the focus of the meeting. Therefore, Prototype A was ultimately selected as the AI pet prototype.

✅ Final choice: Prototype A

Survey Data Results

Product Description Clear

85%

Stability Excellence

90%

Module Usage

80%

Most Time-Consuming/Troublesome Links

57.89%

Personal work progress synchronization difficult

31.58%

Screen sharing switching inefficient

10.53%

No difficulties

Evaluation Methods (3 Types)

Most Time-Consuming/Troublesome Links

57.89%

Personal work progress synchronization difficult

31.58%

Screen sharing switching inefficient

10.53%

No difficulties

Project Value

💡

Visualize discussions with real-time whiteboard, mind mapping, AI assistance

🤝

Lower social and technical barriers, support inclusive participation

🔗

Solve remote meeting fragmentation problem

Transform inefficient communication to structured, recordable, synchronizable collaboration process

Implementation & Conclusion

Implementation Outcomes

✅ Implemented complete Web-side collaboration platform
✅ Supports video conferencing, AI Q&A, mind mapping, file sharing, icebreaking interaction
✅ Completed multiple rounds of user testing and interface iterations
✅ Function button optimization: moved to top, leaving more operation space

Conclusion - Summary

X-Thread provides XJTLU students with an anytime, anywhere group collaboration meeting platform, integrating real-time whiteboard, mind mapping, AI assistance, and icebreaking interaction, transforming scattered discussions into a structured synchronization process, significantly improving communication efficiency and participation experience.

Current Limitations

🔑

Campus Single Sign-On (SSO) Integration

We were unable to implement XJTLU email-based account registration, meaning users currently rely on generic login methods rather than the university's official authentication system.

🤖

Custom AI Integration & Specialized Agent

Our planned integration with XJTLU's exclusive campus AI was not completed. In the future, connecting to this service would eliminate the need for users to configure external API keys. We also aim to explore training a specialized AI agent tailored to support student group discussions.

🔒

Security Testing

Comprehensive security audits and penetration testing have not been performed on the platform. As a result, there is a potential risk of user information leakage, which will require further testing and hardening in future iterations.

📊

Post-Meeting Contribution Analytics

The platform currently lacks post-meeting statistics on member participation and automatic contribution calculation. This feature would help students track engagement, distribute work fairly, and provide transparent records for peer evaluation.

Future Optimization Directions

🔑

Campus Account Integration

We will actively communicate with the university's IT department to explore the implementation of XJTLU Single Sign-On (SSO), enabling one-click login with student emails to streamline registration and ensure exclusive access for campus users. We will also refine role-based permission management, allowing instructors to view group discussion records and project progress as needed.

🤖

Advanced AI Integration

We will continue to coordinate with relevant university teams to integrate XJTLU's exclusive campus AI service, eliminating the need for manual API configuration and lowering the barrier to use. Building on this, we plan to train and deploy a specialized AI agent tailored for student group discussions, providing ice-breaking guidance, real-time academic support, opinion organization, and automated meeting summarization.

🔒

Platform Security Enhancement

Conduct comprehensive security audits and penetration testing to address potential vulnerabilities and prevent user information leakage. We will implement HTTPS encryption, data anonymization, and access control mechanisms to strengthen overall platform security.

📊

Post-Meeting Contribution Analytics

Develop a post-meeting participation statistics module to automatically record metrics such as speaking duration, message count, and file contributions. We will also implement an automatic contribution calculation feature that generates visual reports, providing objective data for peer evaluation and instructor assessment.

Beyond these specific improvements, we will continue to iterate on the website based on user feedback and university requirements, ensuring the platform meets the needs of student collaboration at XJTLU.

Tools & References

Design Tools

Figma, Adobe XD, Miro, Sketch

Development Tools

VS Code, Git, GitHub, Node.js

AI Tools

ChatGPT, GitHub Copilot, DALL-E

Project Management

Trello, Notion, Discord, Google Workspace

Academic References

[1] H. Donelan, K. Kear, "Online group projects in higher education: persistent challenges and implications for practice", Accessed: Mar. 24, 2023. Available: https://pmc.ncbi.nlm.nih.gov/articles/PMC10038701/
[2] Y. Qihao et al., "LLM-based Collaborative Agents with Pedagogy-guided Interaction Modeling for Timely Instructive Feedback Generation in Task-oriented Group Discussions", Accessed: 2025. Available: https://www.ijcai.org/proceedings/2025/1108.pdf
[3] Lotfi. F. Z. et al., "Collaborative Concept Maps in Higher Education: Pedagogical Contributions, Cognitive Challenges, and Optimization Strategies for Interactive Visual Learning", Accessed: Mar. 7, 2025. Available: https://files.eric.ed.gov/fulltext/EJ1466585.pdf
[4] A. Sibarani, "Usability and user satisfaction rate evaluation on e learning application from student's perspective using Nielsen usability method", Accessed: Aug. 20, 2021. Available: ResearchGate

Technical Reflection

AI-Assisted Development Process

During the development of this project website, AI tools were systematically integrated into the design and coding workflow. The following documents the prompts used, verification methods applied, and ethical considerations observed throughout the process.

Prompts Used

1. Playful Geometric — Foundation Prompt

A foundational style prompt was used to establish a unified visual language across the entire website, ensuring consistency in color palette, typography, and geometric decorative elements. This provided the cohesive aesthetic backbone upon which all subsequent sections were built.

2. Component-Level Prompts — Targeted Generation

To prevent visual monotony and enrich design quality, we researched and curated refined UI component prompts from professional design websites. These targeted prompts were used to generate specific components — such as bento galleries, timeline, circular carousel, and interactive cards — each with a distinct visual identity while maintaining overall style coherence with the Playful Geometric theme.

3. Manual Refinement — Human Adjustment

After integrating AI-generated code, we performed manual adjustments tailored to actual content needs: resizing component dimensions to accommodate specific photo aspect ratios, fine-tuning animation parameters (duration, easing curves, trigger behavior), optimizing responsive breakpoints for cross-device consistency, and adjusting spacing and layout to ensure natural visual flow across all sections.

Verification Methods

🔍

Code Review

All AI-generated code was reviewed line-by-line before integration to ensure correctness and compatibility with the existing codebase.

🌐

Cross-Browser Testing

Visual rendering was verified across Chrome, Edge, and Firefox to ensure consistent appearance and behavior.

📱

Responsive Validation

Responsive behavior was tested across mobile, tablet, and desktop viewports to ensure consistent layout adaptation.

Functional Testing

Each interactive component — lightbox modals, navigation, gallery carousels — was manually tested for correct behavior.

Ethical Considerations

🤝

Human Oversight Maintained

AI served strictly as a design and coding assistant, not a replacement for human decision-making. All design direction, content strategy, and final approvals remained under full team control.

🧠

Critical Judgment Applied

AI-generated outputs were treated as starting points, requiring human evaluation for appropriateness, accuracy, and quality before adoption. No code was deployed without team understanding and validation.

🎓

Academic Integrity

We ensured the final deliverable reflects genuine team understanding and capability. AI assistance augmented our workflow but did not substitute for the learning outcomes expected from this coursework.

Project Timeline

April 30, 2026

Final Polish and UI/UX Refinement

  • Conducted final performance audit and responsiveness checks
  • Applied micro-adjustments to layout, spacing, and scroll behavior
  • Resolved minor UI issues including element overlapping and navigation responsiveness
  • Project delivery confirmed
April 25, 2026

Deployment and Environment Stabilization

  • Successfully deployed frontend to Render cloud hosting
  • Verified production environment configurations and API endpoints
  • Monitored system stability and load-time performance
April 20, 2026

Comprehensive System Testing

  • Executed full system test suite covering all functional modules
  • Validated end-to-end user flows in integrated staging environment
  • Verified cross-browser compatibility and responsiveness across mobile/desktop
April 13, 2026

Phase Summary and Deployment Planning

  • Systematically summarized project progress in development, testing, and deliverable production
  • Confirmed on-time completion of function implementation, test validation, and poster production
  • Developed a plan for web deployment and public release in next phase
April 12, 2026

Document Standardization and Acceptance Preparation

  • Standardized academic documents covering requirements, design, testing, and summary sections
  • Completed internal acceptance and deliverable confirmation in local environment
April 11, 2026

Full System Regression Validation

  • Ran comprehensive regression testing to ensure stable performance with no regressions
  • Validated collaboration workflows, interface rendering, and data persistence effectiveness
April 10, 2026

Poster Finalization and Deliverable Integration

  • Completed visual optimization, content proofreading, and finalization of academic poster
  • Compiled project resources, code, and documentation into a standardized deliverable package
April 9, 2026

Academic Poster Design and Content Development

  • Established structure for X-Thread achievement poster in line with academic poster standards
  • Organized core sections including research objectives, methodology, implementation, and conclusions
April 8, 2026

Functional Verification and Compliance Checking

  • Executed full regression testing to confirm system meets module design requirements
  • Performed compliance verification against CPT208 design specifications and methodologies
  • Issued a phased test report
April 7, 2026

Defect Resolution and Test Case Expansion

  • Identified and fixed functional defects exposed during integration testing
  • Expanded test cases for edge and exception scenarios to improve test coverage
April 6, 2026

Local Integration and System Testing

  • Deployed a complete local testing environment and conducted end-to-end integration testing
  • Verified functional integrity, data consistency, and concurrent reliability across full workflows
  • Logged and prioritized defects by severity
April 5, 2026

Public Holiday (Day Off)

Suspended all research and development activities; no development or testing tasks

April 4, 2026

Functional Closure and Unit Testing

  • Completed development of all module-specified functions to achieve full requirement coverage
  • Performed unit testing to verify functional correctness of individual components
  • Conducted code review and version archiving
April 3, 2026

User Experience Optimization and Robustness Enhancement

  • Optimized interaction flows and interface feedback mechanisms to improve system usability
  • Fixed logical anomalies under boundary conditions to strengthen system stability
April 2, 2026

Functional Iteration and Modular Integration

  • Advanced coding of key functions and refined business logic
  • Carried out front-end and back-end integration to validate data transmission and state consistency
April 1, 2026

Implementation of Core Functional Modules

  • Finished system infrastructure setup and environment configuration
  • Implemented core functions including thread management, content editing, and real-time collaboration
March 31, 2026

System Architecture and Technical Scheme Design

  • Completed overall architecture design and modular decomposition of X-Thread
  • Formulated interaction workflows, data models, and interface specifications
  • Produced technical design documents to ensure development consistency
March 30, 2026

Requirement Analysis and Research Framework Establishment

  • Conducted requirement elicitation and analysis for X-Thread system following Human-Centric Computing paradigm
  • Defined design methodology, technical roadmap, and evaluation metrics
March 29, 2026

First Draft Design

Designed first draft of the poster and completed the overall framework

March 18, 2026

Project Homepage Deployment

Completed the project homepage with team member information and deployed to GitHub Pages

March 17, 2026

Task Division and Theme Confirmation

Divided tasks among 4 team members and confirmed the project theme

March 15, 2026

Project Kickoff

Created GitHub repository and initialized the project structure