Misinformation Investigate - UI

Demo link - https://www.youtube.com/watch?v=mvKfc1Jfy8Q

---

The user interface which will be used by users to start live feed processing using m3u8 links. Users will be able to check any fake news which have been detected.

Instructions to run

Serivices/Apps required

• The User interface
• Live Feed API
• Speech to text API
• Prediction API
• Fact Checking API
• Flink stream processing pipeline
• Kafka broker

Kafka

---

Requirements

1. Scala 2.13.14
2. Apache Kafka 2.13-3.9.0

We will run the Kafka with KRaft mode

• Format the kraft metadata directory using the command -

  bin/kafka-storage.sh format -t <unique_cluster_id> -c config/kraft/server.properties

• Edit the config/kraft/server.properties file

• Add proper listeners and controller ip addresses

• Create kafka topics. Example -

  bin/kafka-topics.sh --bootstrap-server <ip_addr:9092> --create --topic <topic_name> --partitions <p_count> --replication-factor <rep_count>

• Run the kafka server - bin/kafka-server-start.sh config/kraft/server.properties

User Interface

---

Requirements

1. Node and npm

Clone this repo

• cd into repo folder
• Run - npm install
• Install the dependencies which are asked
• Run - npm run dev

Live Feed API

---

Requirements

1. Python 3.12

Clone the repo - https://github.com/abhimessi16/Misinformation-Investigate-Feed-API

• cd into repo folder
• Create a virtual environment - recommended
• Run - pip install -r requirements.txt
• Create a .env file using .envExample. Add the kafka broker urls.
• Change the kafka topic name in params file if necessary
• Run - fastapi run app/main.py (or run using uvicorn on different port)

Speech to Text API

---

Requirements

1. Python 3.12

Clone the repo - https://github.com/abhimessi16/MI_Processing_STT

• cd into repo folder
• Create a virtual environment - recommended
• Run - pip install -r requirements.txt
• Run - fastapi run app/main.py (or run using uvicorn on different port)

Prediction API

---

Requirements

1. Python 3.10.12

Clone the repo - https://github.com/abhimessi16/MI_Fake_News_Prediction

• cd into repo folder
• Create a virtual environment - recommended
• Download and add the model and tokenizer files from - https://drive.google.com/drive/folders/1LSrSNxbLMT5yubg_tghWe1suH8qvGkmE?usp=sharing
• If using own models then update the main, params and utils files accordingly
• Run - pip install -r requirements.txt
• Run - fastapi run app/main.py (or run using uvicorn on different port)

Fact Checking API

---

Requirements

1. Python 3.12

Clone the repo - https://github.com/abhimessi16/MI_Processing_Check

• cd into repo folder
• Create a virtual environment - recommended
• Run - pip install -r requirements.txt
• Create .env file using .envExample. Add your Google API key. This API key must have access to the Google Fact Check Tools API
• Run - fastapi run app/main.py (or run using uvicorn on different port)

---

Can use the same virtual environment for all python APIs except Prediction API.

---

Flink Streaming

Requirements

1. Java 11
2. Maven

Clone the repo - https://github.com/abhimessi16/mi-processing

• cd into repo folder
• Create application.properties using example.properties file
• Enter all the relevant details in properties file
• If running using jar file then make sure flink dependencies are built.
  • Follow - https://nightlies.apache.org/flink/flink-docs-release-1.4/start/building.html
  • Run - mvn clean install
  • Run the jar file
• If running as a project
  • make sure all dependencies are built
  • Run DataStreamJob's main method