Project Stages


Milestone 1: Planning

Our Team
wrapkit
Inês Ferreira
Product Owner
wrapkit
Guilherme Lopes
Team Manager
wrapkit
Simão Antunes
Architect
wrapkit
Pedro Matos
DevOps
wrapkit
Diogo Gaitas
QA Tester
wrapkit
Mário Antunes
Advisor
Concept
...
Situation

There are a wide variety of sites, portals and communication services (e.g. discord, slack, etc) that contain different types of information.

...
Problem

There are too many different sources, the information is widely spread.


...
Solution

Creating an agent that answers any question about the department correctly and in a natural way.

wrapkit
Core Idea

DetiBot helps you getting relevant information with natural interactions.

This project aims to develop a DETI Bot agent that leverages both Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) to monitor and index all relevant data sources pertaining to departmental activities. The bot will offer diverse interfaces for the community to seamlessly interact with this information.

Who is it for?
STUDENTS

To know about events, room locations, schedules and services. 

PROFESSORS

To know of events, educational files, administrative contacts, research groups, and projects. 

STAFF

For easy access to different DETI contacts. 

VISITORS

To view DETI presentation, demos, projects, and events open to the public. 

Key Functionalities

Information gathering from different DETI websites/documents.

Optimized information indexing.

Unified API with access to multiple front-end possibilities (web page, IM bots).

SWOT Analysis
S
Strenghts

It provides a natural interaction with its users. Encompasses lots of information

W
Weaknesses

Has to deal with fragmented information. Depends heavily on other technologies. Has to prevent inadequate answers from agent

O
Opportunities

AI market in extention. Using innovative technology to boost/rejuvenate DETI. Taking advantage of the lack of centralized information.

T
Threats

Difficulty acessing/updating/compartmentalizing information. Difficulty in ensuring infrastructure.

Project Calendar

Milestone 2: Architecture

Personas
Scenario

Joana wants to know when her easter holidays start and end


"Quando são as férias da Páscoa?"

"As férias da Páscoa na Universidade de Aveiro irão decorrer entre 25 e 31 de Abril"

wrappixel kit
Joana Mota

Joana is a 22 year old DETI student that lives in Aveiro.

Scenario

Francisco is interested in studying LEI in Aveiro. He wants to know what subjects he'll get to know in his first year.


"Que cadeiras irei frequentar no primeiro ano do curso Engenharia informática em Aveiro?"

"No primeiro ano de Engenharia Informática irá frequentar as seguintes cadeiras: Fundamentos da programação, Introdução às Tecnologias Web, ..."

wrappixel kit
Francisco Vale

Francisco is a 18 year old Highschool Student that lives in Porto.

Functional Requirements


Functional requirements specify the actions or tasks a system must perform to meet user needs.

Select language between portuguese or english
Select trending querys
Ask questions by text or voice
Receive answers by text or voice
View chat history
Rate answer accuracy

Non-Functional Requirements


Non-functional requirements specify criteria that characterize the operation of a system rather than specific behaviors.

Availability: Error Handling and Fault tolerance
Performance: Scalability and Response Time
Usability: User-friendly UI and Accessibility
Maintainability: Modularity and Documentation

Actors and Use Cases

System Architecture

wrapkit

What is qdrant ?

qdrant is a vector similarity search engine that provides a production-ready service with a convenient API to store, search and manage points (i.e. vectors) with an additional payload



Domain Model

Why do we need a Vector Database
Efficient storage and indexing of high-dimensional data.
Ability to handle large-scale datasets with billions of data points.
Ability to handle vectors derived from complex data types such as natural language text
Improved performance and reduced latency in machine learning and AI applications. Reduced development and deployment time and costs compared to building a costum solution

Deployment Diagram


Milestone 3: Prototype

Future Work

,

Milestone 4: Final Product

Future Work