Final Project

1. Requirements

● Functionality:
- Build a system for storing data (e.g., data warehouse, NoSQL databases,
hadoop big file system, …)
- Collect the data and prepare data for analysis
- Build a dashboard for visualizing data that supports understanding the
overview of data
- Data analysis: providing statistical analysis and applying advanced machine
learning/AI models for mining hidden knowledge/structure of the data that is
useful for real-world business.
● Problems: Select one of the problems in Section 5

2. Registration
Fill in the following Google docs file. Note that up to 4 groups are allowed to register
the same problem. A group can have up to 4 members.

https://docs.google.com/spreadsheets/d/1_0mP2Sdz2a3cjGKiZVHK4Il6I3Beqfyu/edit?usp=s
haring&ouid=106996255610935127545&rtpof=true&sd=true

Deadline: Oct 29, 2024

3. Checkpoints

There will be a few checkpoints. On those days, each group must present the work in
progress. You must pass all the checkpoints before having the final presentation.
The schedule for checkpoints will be announced later.

4. Final submission and presentation.

You are required to submit the report, source code, and presentation slides. The
submission link will be announced later. On the presentation day, all members of the
group must attend the class and each member is recommended to present his/her own
part.
Deadline: Dec 17, 2024

5. Problems
5.1 Evaluation of Twitter Account Quality for KOLs in the Web3 Space
In Web3 marketing, Key Opinion Leaders (KOLs) on social media platforms like
Twitter are crucial for driving trends, building communities, and shaping opinions. This
project aims to develop a robust framework to assess the quality of Twitter accounts
operated by KOLs in the Web3 space, providing insights into their influence,
effectiveness, engagement, and credibility.

Requirements:

● Identification of KOL Twitter Accounts: Compile a comprehensive list of

Twitter accounts belonging to KOLs being actively involved in the Web3
space.
● Data Collection Workflows: Develop and implement automated workflows
for collecting relevant data from Twitter.
● Deployment of Machine Learning, AI, and Deep Learning Models: Utilize
advanced techniques in machine learning, artificial intelligence, and deep
learning to evaluate the quality of each KOL's Twitter account.

5.2 Evaluation of Reputation Score and Trending Projects for NFTs, Tokens, and
Web3 Projects on Social Media

In the rapidly evolving Web3 ecosystem, the reputation and popularity of NFTs, tokens,
and projects play a significant role in driving adoption and investment. This project
evaluates the reputation scores and trending status of various NFTs, tokens, and projects
based on their social media presence and community engagement.

Requirements:

● Data Collection from Web3-related Social Media Platforms: Implement

processes to gather comprehensive data from various social media platforms
where discussions about NFTs, tokens, and Web3 projects take place.
● Integration of Data from Multiple Platforms: Combine and integrate the
collected data from different social media sources to create a unified dataset.
● Evaluation of Popularity and Reputation in Communities: Analyze the
integrated data and apply ML or DL models to assess the popularity and
reputation of specific NFTs, tokens, and projects.

5.3 Twitter KOLs and Telegram Contributors Recommendation System

To drive growth and engagement in Web3 projects, identifying and connecting with the
right influencers and potential contributors is essential. This project aims to suggest and
link Key Opinion Leaders (KOLs) and potential collaborators from platforms like
Twitter and Telegram to specific projects, helping to boost visibility and foster
partnerships.

Requirements:

● Data Collection from Twitter, Discord, and Telegram: Implement

workflows to gather relevant data from social media platforms such as Twitter,
Discord, and Telegram.
● Labeling Accounts as KOLs or Potential Contributors
● Building a Recommendation System for KOLs and Contributors: Develop
a recommendation system to suggest KOLs and potential collaborators for
Centic's list of projects.

5.4 Post Suggestions for Web3 Influencers and Marketers

Creating engaging and informative content is essential for Web3 influencers and
marketers to build credibility and grow their audience. This project focuses on
generating impactful post suggestions based on blockchain data analysis and market
trends, helping influencers and marketers maintain a strong presence on platforms like
Twitter and Telegram.

Requirements:

● Blockchain Data Analysis Provided by Centic: Use the blockchain data

provided by Centic to perform an in-depth analysis of market trends, user
behavior, token movements, and project performance.
● Utilize NLP and LLM Models for Automated Post Generation: Employ
Natural Language Processing (NLP) and Large Language Models (LLMs) to
automatically generate well-structured posts for Twitter, Telegram, and other
platforms.
● Integrate Analysis with Post Generation and Suggest Content for
Influencers and Marketers: Combine the analytical insights with automated
content generation to create comprehensive and insightful posts.

5.5 Tracking and Evaluating the Effectiveness of KOLs and Contributors with
Corresponding Projects

To ensure that Key Opinion Leaders (KOLs) and contributors are effectively supporting
specific Web3 projects, this project focuses on tracking and evaluating their
performance based on social media engagement. The goal is to assess how well these
influencers and collaborators promote and contribute to the growth of Centic’s projects.

Requirements:

● Data Collection from Twitter and Telegram: Implement systems to collect

relevant data from Twitter, Telegram, and other social media.
● Mapping KOLs and Contributors to Projects Provided by Centic: Match
KOLs and contributors to the specific projects they are promoting or associated
with.
● Evaluating the Effectiveness of Tweets and User Interactions with
Projects: Analyze the performance of KOLs and contributors by evaluating the
effectiveness of their tweets, posts, and interactions with users regarding the
projects.

5.6 Cash Flow Analysis and Investment/Trading Strategy Recommendations on

DEXs

In the decentralized finance (DeFi) space, understanding cash flow patterns and
developing effective trading strategies are key to success. This project focuses on
analyzing cash flow and providing investment and trading strategy recommendations,
leveraging data from Centic and additional sources in the Web3 ecosystem.

Requirements:

● Utilize Centic Data and Collect Additional Web3 Data: Analyze the
financial and transactional data provided by Centic, and supplement it with
additional data collected from decentralized exchanges (DEXs) and relevant
blockchain platforms.
● Analyze Cash Flow Patterns on DEXs: Perform in-depth analysis of cash
flow trends on DEXs by tracking large transactions, liquidity movements, and
the behavior of prominent wallets.
● Recommend Investment and Trading Strategies: Based on the cash flow
analysis and other relevant data, provide strategic recommendations for
investment and trading.

5.7 Collaboration Suggestions Between Projects

To foster growth and innovation within the Web3 space, collaboration between similar
projects can be highly beneficial. This project focuses on identifying opportunities for
partnerships between projects by analyzing their data and communities, both from
Web3 and Web2 platforms.
Requirements:

● Collect and Use Data from Web3, Web2, or Both: Gather data from Web3
sources (such as blockchain data and decentralized apps) as well as Web2
platforms (social media, websites, user communities) for various projects.
● Evaluate the Similarities Between Projects and Their Communities:
Analyze the data to assess the degree of similarity between different projects.
● Suggest Collaborations Between Projects with High Similarity: Based on
the analysis, recommend collaboration opportunities between projects that
share significant similarities.

5.8 Automated Chatbot System Built from Website Content

Description:
This system automatically analyzes, classifies, and extracts information from any given
website. Based on this data, a vector database is created, allowing for the rapid
development of a question-answering chatbot. The goal is to generate a chatbot version
that can efficiently respond to user inquiries in a short time frame.

Related technique:
- Language models
- Retrieval-Augmented Generation
- Prompt engineering
- Web scraping
- Vector database

5.9 Recommendation system for a chatbot in the blockchain domain

Description:
Building a recommendation system for suggesting follow-up queries to users based on
asset information, query history, and the system’s goals, such as encouraging users to
perform transactions or take actions in a certain direction. This aims to balance between
user experience and the system’s profit/revenue goals.

Related technique:
- Knowledge graph
- Multi-turn recommendation
- Sequential conversational target-oriented recommend system

5.10 Automated system for building real-time entity data in the blockchain domain
Description:
Entities in the blockchain domain, such as tokens, dApps, etc., are vast in number and
are continuously added or updated daily. Currently, there is a significant lack of quality
datasets related to blockchain in general and blockchain entities in particular, which are
necessary for natural language processing tasks. Therefore, developing an automated
system that constructs diverse datasets about entities specifically, and other related data
in general, is essential to support the implementation of various additional tasks.
Related technique:
- Language models
- Multi-agent LLM system
- NoSQL and SQL database systems
- Data Integration and Processing