tweetcapture

GCP tweet capture and sentiment analysis

This demo uses the following components of GCP: kubernetes, docker, pubsub, Natural Language (sentiment analysis), BigQuery, and Data Studio

Setup

Clone the example repo

If you haven't already, clone the Github repository for this demo:

git clone https://github.com/NeudesicGCP/GCPTweetCapture.git

Set project

gcloud config set project neu-ncaa

Changing topics

The topics that are being captured are found in the file neu-ncaa/tweetcapture/pubsub/tweet-capture-docker-image/twitter-to-pubsub.py

Edit the list of topics using your favorite editor

nano twitter-to-pubsub.py

Create the docker image using cloud build - (if topics are changed)

(from /neu/tweetcapture/pubsub/tweetcapture-docker-image folder)

gcloud builds submit --tag gcr.io/neu-ncaa/tweetcapture-image .

Capturing tweets

Kubernetes cluster create (cloud shell)

gcloud beta container --project "neu-ncaa" clusters create "ncaatweet-cluster" --zone "us-central1-a" --service-account "neu-ncaa-tweet-serviceaccount@neu-ncaa.iam.gserviceaccount.com"

gcloud container clusters get-credentials ncaatweet-cluster --zone us-central1-a --project neu-ncaa

Create the pods

(from /neu-ncaa/tweetcapture/pubsub folder)

kubectl create -f twitter-stream.yaml

kubectl create -f bigquery-controller.yaml

Check the pods

kubectl get pods

Stoping the tweet capture

kubectl delete deployment -l "name in (twitter-stream, bigquery-controller)"

###################################################################################

Sentiment Analysis using Google Natural Language API

(must be run from a virtual environment using CloudShell)

pip install virtualenv

virtualenv twitter

source twitter/bin/activate

twitter/bin/pip install google-cloud-language

pip install google-cloud-bigquery

Process tweet sentiment

(from /neu-ncaa/tweetcapture/sentiment folder)

run during and after the tweet capture

python bigquery-sentiment-analysis-naturallanguage.py ncaa_tweets

###################################################################################

Visualization - Data Studio

This visualization in Google Data Studio uses several views in BigQuery that pull from the tweet and tweetsentiment tables

https://datastudio.google.com/open/1m4S2Hqn1BCi7ql_2sfCEzz6H7x0oiAEY

###################################################################################

Cleanup

Delete kubernetes cluster

gcloud beta container --project "neu-ncaa" clusters delete "ncaatweet-cluster" --zone "us-central1-a"