An opinion mining platform, modular and open sourced

A PhD project, being developed by Miguel Fernandes for the University of Vigo, under the supervision of Ana Fernández Vilas and Rebeca Díaz Redondo.

Main Features

Tweets Extraction

Directly from twitter, with your own API Keys. Tweets are extracted using the Search API, and in the near future, the Stream API for live events.


Texts are cleaned, trimmed, separated, tokenized, lemmatized and processed, using several Natural Language Processing (NLP) techniques available.

Sentiment Analysis

Do this message express happiness? sadness? is the customer satisfied? did our decision had a bad impact on our users? for how long? Questions aimed to be answered by the platform.


Convert data into information. Evolution graphs, comparison charts, exportable tables. All on your own desktop browser, smartphone or tablet. One click away.

Frequently Asked Questions

Is it ready?

No, this is a work in progress, but we will be releasing some nightlies soon. We are working at a fast pace developing the platform, but as a work in progress, you should be aware that bugs will be all over the place. Features will be added on a weekly basis, before we enter into the feature freeze phase.

I want to try it anyway. Can I?

Sure!. You will need a server with internet connection and docker installed, and everything else will be taken care of. Be sure to follow the instructions on our github site, and you will get a MongoDB, an Eureka registry and the full marble platform using a single command from the shell.

Where is the documentation?

You can find the documentation in the project's Wiki, and in the main README file of the project.

Is everything developed by you?

Not at all. We are using libraries and software developed by other people, and we will give credit to them appropiately in our third party section (still to be developed). If you think we are using some of your software inappropriately, please let us know.

How is this project licensed?

We are releasing all the software developed for this project under the Apache license 2.0. Click here for more information about it.

* I'm trying to access the demo but it seems down.

Please open an issue on github to let me know.