Documentation

Architecture of the pipeline

create nlp-pipeline docker container

1. Change to the directory with the dockerfile
2. run docker build . -t nlp-pipeline
3. start docker container docker run -p 127.0.0.1:5000:5000 --name nlp-engine nlp-engine

pull docker image from dockerhub and start

1. download docker image docker pull modius22/nlp-pipeline
2. start docker container docker run -p 127.0.0.1:5000:5000 --name nlp-pipeline modius22/nlp-pipeline

start api from terminal

1. change in the src directory
2. run python api.py

example run of each function

standard function

create new project: curl -i 'http://127.0.0.1:5000/nlp/create_project?project_name=project_a'

get list of projects: curl -i 'http://127.0.0.1:5000/nlp/get_project'

get list of all files of an project: curl -i 'http://127.0.0.1:5000/nlp/get_project_files?project_name=project_a'

remove existing project: curl -i 'http://127.0.0.1:5000/nlp/delete_project?project_name=project_a'

nlp function

load data to new project: curl -i 'http://127.0.0.1:5000/nlp/project_a/load?text=text&sentiment=airline_sentiment' -F file=@./Tweets.csv -X POST -H 'enctype:multipart/form-data ; Content-Type:multipart/form-data'

explore data: curl -i 'http://127.0.0.1:5000/nlp/project_a/exploartion'

clean data: curl -i 'http://127.0.0.1:5000/nlp/project_a/clean'

vectorize data: curl -i 'http://127.0.0.1:5000/nlp/project_a/vectorize'

learn model: curl -i 'http://127.0.0.1:5000/nlp/project_a/model?algorithm=LR'

predict: curl 'http://127.0.0.1:5000/nlp/project_a/predict?algorithm=LR&text=told%20work%20joke%20fail' -X POST -v

learn model MNB curl -i 'http://127.0.0.1:5000/nlp/project_a/model?algorithm=MNB'

predict MNB curl 'http://127.0.0.1:5000/nlp/project_a/predict?algorithm=MNB&text=told%20work%20joke%20fail' -X POST -v

known problems

macOS curl on terminal "zsh: no matches found:"

On macOS the curl command may cause problems. The solution is to set the URL in '.

e.g. curl -i 'http://127.0.0.1:5000/nlp/create_project?project_name=project_a'

no access to the working folder

This problem can occur if you start the API from a development environment. In this case the working directory for the development environment must be set to the ./src folder.