Sentiment Analysis using Apache Flink¶
This project runs a Sentiment Analysis application with Apache Flink Stream API.
You can interactively type in sentences to simulate stream data, and Flink will process and show the result in the terminal.
Step to run¶
Firstly you need to create a simple text server by doing the following in the terminal:
Linux/Mac¶
nc -l 9000
Windows¶
nc -l -p 9000
After that you can open another terminal in the project and run the follows:
./gradlew run
On Windows use gradlew.bat
instead.
Try to type in anything from your text server:
$ nc -l 9000
I love you!
result shows on Flink application:
Loading: 100% |████████████████████████████████████████|
[Flat Map (10/12)] INFO ai.djl.pytorch.engine.PtEngine - Number of inter-op threads is 6
[Flat Map (10/12)] INFO ai.djl.pytorch.engine.PtEngine - Number of intra-op threads is 1
[
{"class": "Positive", "probability": 0.99870}
{"class": "Negative", "probability": 0.00129}
]