Sentiment Analysis Spring MVC

A simple Spring Boot MVC application that analyzes user-submitted reviews for sentiment (positive/negative), persists results, and allows users to provide feedback for retraining.

API & Frontend Overview

Path	Method	Description
/	GET	Renders the homepage with a review form, version info, and past inferences.
/	POST	Submits a review for sentiment inference. Result is stored and shown on page.
/train	POST	Submits corrections (disagreed predictions). Flips sentiment and stores feedback.
/train	GET	Returns all user feedback entries as JSON for retraining or inspection.
/actuator/prometheus	GET	Exposes Prometheus-formatted metrics via Micrometer for observability.

Feature Summary

Feature	Details
Review Submission	Users enter a review via a form. It is sent to the model API (/predict via MODEL_URL) and classified.
Result Display	All prior predictions are shown on the homepage, along with version info and a feedback checkbox.
Feedback Submission	Users can select incorrect predictions and submit feedback. The system stores flipped sentiments.
Feedback API	Accessible at /train (GET) as JSON. Useful for training data collection.
Metrics	Inference time, failure counts, and stored review counts are tracked and exposed via Prometheus.

Getting Started

Prerequisites

• Java 17+
• Maven 3.6+
• A running sentiment model REST endpoint at /predict

Installation

With Docker

1. Download app package from the repository or from ghcr.io/remla25-team6/app
2. Run it with docker

   docker run -p 8080:8080 -e MODEL_URL='<MODEL_URL>' app

Without Docker

1. Clone the repository:

   git clone https://github.com/remla25-team6/app.git
   cd app

2. Configure the following environment variable:

   MODEL_URL=<YOUR MODEL URL>

3. Authenticate Maven to GitHub repository via PAT:

   // in /.m2/settings.xml
   <settings>
     <servers>
       <server>
         <id>github</id>
         <username>YOUR USERNAME</username>        <!-- your GitHub username -->
         <password>YOUR PAT</password>    <!-- your PAT  -->
       </server>
     </servers>
   </settings>
   

4. Build the project with Maven:

   mvn clean package

Monitoring

Micrometer metrics are available at /actuator/prometheus and can be scraped by Prometheus. Custom metrics are configured via WebMetrics.java and MetricsConfig.java.

CI/CD

This project uses GitHub Actions to automate testing and versioned releases.

CI

Every commit and pull request triggers the cicd.yml workflow, which:

• Builds the application using Maven
• Runs unit tests
• Verifies build and integration success

Release

To publish an official release:

1. Ensure all changes are committed and pushed to a release-ready branch.
2. Tag the commit with a semantic version like v0.1.0 and push:

git tag v0.1.0
git push origin v0.1.0

3. This triggers the release.yml workflow, which:

• Builds the application
• Publishes a GitHub Release with the JAR file attached

Pre-release

To publish a pre-release:

1. Push a commit to the main branch (e.g. through a pull request to main).
2. The prerelease.yml workflow runs automatically on each commit to main:

• Builds and packages the app
• Publishes a pre-release on GitHub with a timestamped version (e.g., v0.1.0-pre.20250625.185600)