project-proposal-2025

[AI Chatbot Integration for Blackboard]

Abstract

When I have a question about a tutorial or general queries, I must manually navigate between Blackboard and EdDiscussion, typing keywords to find answers. This requires multiple steps and mouse clicks. The separation of these platforms makes it even harder to locate relevant resources, especially when tutors reference materials on EdDiscussion, forcing students to search for them manually on Blackboard. An AI chatbot that integrates both platforms would simplify this process, allowing students to quickly find discussions, resources, and answers without switching between systems. This would improve information retrieval efficiency, reduce search time, and provide a seamless user experience.

Author

Name: Theodore Kim

Student number: 47476749

Functionality

• AI Chatbot: Users will interact with a chatbot embedded in the Blackboard homepage. It will respond to simple questions related to course content, assignments, discussions, and relevant resources.
• Seamless Integration: The chatbot will pull relevant answers from both Blackboard resources and EdDiscussion threads, including course materials and discussion posts. Chatbots are common on many websites, so integration should be smooth and unobtrusive.
• Search and Recommendation: The system will allow students to ask questions or enter keywords to quickly locate discussions and relevant resources on both platforms.
• Smart Suggestions: Based on the query, the system will suggest related materials in Blackboard (e.g., lecture notes, reading materials) and past discussions in EdDiscussion.

Scope

Minimum Viable Product (MVP)

The MVP will include an AI chatbot that:

• Answers basic course-related questions.
• Displays or redirects users to relevant resources from Blackboard.
• Provides references to discussions in EdDiscussion based on keywords.
• Uses a Retrieval-Augmented Generation (RAG) structure to improve response accuracy while limiting results to publicly available information.
• Maintains privacy and role-based access permissions to data, may need to develop a way to sort both BlackBoard and EdDiscussion Board’s data during preprocessing of the vectorization, in a format that can be filtered by the software

Relevant Notes

Please do some reading into RAG and how it works, as it is a necessary structure to connect a Large Language Model (LLM) with the data found in BlackBoard and EdDiscussion board. For the MVP, you could scrape data from your personal BlackBoard and EdDiscussion Boards and place them in a databases or local directory. The goal is to make use of vectorised forms of both the BlackBoard directory and the Q&A, which may not necessarily be data found in one singular database due to both sites being independent in practice, but for MVP assume all data is coming from one database/directory. For privacy reasons, if a post on EdDiscussion Board is private, create a filtering system that does not include private posts.

Quality Attributes

QA Trade-offs and Priorities

• Interoperability vs. Modularity: Interoperability is prioritised over modularity because the system must function seamlessly across two separate platforms: Blackboard and EdDiscussion. Ensuring smooth communication and data exchange between these platforms is critical for the chatbot to deliver its intended functionality. If modularity is prioritised, the system could become disjointed, leading to integration issues where data might not flow properly between the components. While modularity helps with future extensions and maintainability, it’s less important than interoperability, as a system that doesn’t integrate well across platforms will ultimately fail to meet user needs.

• Scalability vs. Reliability: Scalability is prioritised because the system must be able to accommodate a growing number of users without compromising performance. A scalable system can handle an increasing load efficiently, ensuring that the chatbot remains responsive even as more students interact with it concurrently. While reliability is important for system stability, focusing solely on reliability could limit the system’s ability to scale effectively. A reliable system that can’t scale would struggle under heavy usage, causing delays and failures. Scalability ensures long-term viability, whereas a system that is only reliable at smaller scales may encounter significant issues as the user base expands.

Security

The system must maintain normal operations even when subjected to potential attacks. Since EdDiscussion and Blackboard are separate services with different permission structures, the chatbot must respect user roles and access controls when handling data from both platforms.

• Measurement: Implement role-based access control (RBAC) to ensure appropriate permissions. Security will be assessed through penetration testing and logging unauthorised access attempts.

  Interoperability

  The chatbot must be able to retrieve, process, and display information from both Blackboard and EdDiscussion without conflicts. It must ensure that data shared across platforms maintains integrity and complies with access rules.

• Measurement: Successful retrieval of data from both platforms without permission errors or data conflicts. Integration testing with mock student accounts will verify smooth data exchange.

  Scalability

  The chatbot should handle increasing user loads without significant performance degradation. As more students use the system, it should maintain response times within an acceptable range.

• Measurement: Load testing will simulate concurrent users and ensure response times remain below a specified threshold (e.g., <2 seconds for 95% of queries).

Evaluation

Security

To assess the system’s ability to handle potential security threats, Role-Based Access Control (RBAC) Testing will be conducted. This method focuses on ensuring that users are only able to access data appropriate to their role, such as students being restricted from accessing instructor-only materials. The testing will verify that the system’s permissions are correctly configured and enforced, ensuring that users cannot bypass security measures to gain unauthorized access. The system will also be evaluated for vulnerabilities in data transmission, such as interception of sensitive data.

Success Criteria:

• No unauthorized access to restricted data, as confirmed through role-based access testing.
• Security mechanisms are effective in ensuring that data is only accessible by those with appropriate permissions.
• The system is resilient against potential threats, such as man-in-the-middle attacks or unauthorized data exposure.

Interoperability

To ensure smooth integration between Blackboard and EdDiscussion, the following evaluations will be conducted:

• API and Data Flow Testing: Verify that the chatbot correctly retrieves and displays information from both platforms without permission conflicts.
• Mock User Testing: Use test student and instructor accounts to ensure the chatbot respects role-based permissions across different platforms.
• Error Handling Checks: Test scenarios where data retrieval fails and ensure the chatbot provides appropriate fallback messages.

Success Criteria:

• The chatbot correctly retrieves data from both platforms without errors.
• No unauthorized data is accessed due to permission mismatches.
• The chatbot provides relevant responses even when API failures occur.

Scalability

To evaluate system performance under different user loads, the following assessments will be conducted:

• Load Testing: Simulate concurrent users (e.g., 100, 500, and 1,000 users) to measure response times and system stability.
• Stress Testing: Push the chatbot beyond expected user loads to identify performance bottlenecks.
• Performance Monitoring: Track response times and system resource usage under varying loads.

Success Criteria:

• The chatbot maintains a response time of <2 seconds for 95% of queries under normal loads.
• The system remains stable under stress tests, with minimal performance degradation.
• Resource utilisation stays within acceptable limits without crashes or excessive lag.

Word Count: 1187 Words