majercakdavid / stock-sentiment-analysis Goto Github PK

Currently, there is only one instance for sentiment analysis running at a time. However, this is not enough for assessing all the tweets. Therefore, it is required to run multiple instances at the same time depending on the current load.

Currently, there is no description about the project. Thus, it is necessary to write one.

Currently, the sentiment analysis job is communicating with other through Kafka channels. However, it might be better to not pollute it twice with almost same data and use gRPC instead for communication.

Currently, there are docker compose files for different stacks. It is essential to unify them to better describe the dependencies between the containers. As a result there will be one docker-compose file which will deploy the application without issues like not yet running required service.

Create a web UI which will show the user currently tracked terms and moving averages for stock quotes as well as a currently monitored sentiment and moving averages for them. Moreover, user should be able to pause and restart processes at any time from within the UI as well as set the tracking terms and moving averages.

Right now it is necessary to always rewrite the domain to the localhost. In order to improve developer experience and productivity it is handy to do it automatically based on the environment in which a particular app is running.

Example:
Docker | Local
broker:29092 | localhost:9092

Currently, on restart of a service - or creating a new instance of the same one, it consumes all messages produced so far for a given channel. This should be fixed and the instance should consume only not consumed messages yet.

Currently, the docker containers are simply killed and produce error logs. To get rid of them it is necessary to make the jobs react to the SIGTERM signal and terminate gracefully instead of forcefully.