SentimentAnalysis

This project was created as part of Alfresco #DevCon2019 and aims to create a micro service that is able to analyise the sentiment of a content. As part of this project we will compare the different NLP frameworks and try to improve on the accuracy of the prediction.

Tasks

• Project setup.
• Data preparation.
• NLP Sentiment Analysis interface.
• StandfordNLP implementation.
• OpenNLP implementation.
• SpringBoot setup to expose Rest-API endpoints
• Unit test.

Stretch Tasks

• Integrate the service into Alfresco, create a rule that invokes the Sentiment Analysis API.
• Integrate with APS to perform a process based on the Sentiment result.

Setup

Change into the sentiment-analysis folder and perform the following:

Building The Application

mvn clean install

Running The Application

java -jar target/sentiment-analysis-1.0.0.jar To view the page: http://localhost:8080/

Docker Build

docker build -t SentimentAnalysis .

Running with Docker

docker run SentimentAnalysis To view the page: http://localhost:8080/

REST API

Get Sentiment Analysis ranking by sentence

Request

$ curl -X POST \
  http://localhost:8080/analyze/v1/text \
  -H 'Content-Type: text/plain' \
  -d Text coming from Alfresco Content

Response

{
    "ranking": {
        "positive": 3,
        "negative": 4,
        "neutral": 3
    }
}