MixedEmotions – the Open Source platform* for emotion extraction
The MixedEmotions platform is a Big Data Toolbox for multilingual and multimodal emotion extraction and analysis (*be aware the plattform is in a beta stadium and still under development). It can extract emotions from text, audio and video. However, it also has many other capabilities, such as sentiment analysis, social network analysis and knowledge graphs visualization among others. Take me directly to the MixedEmotion’s Platform.
In short: What can I do with it?
The platform is a toolbox which anyone can download and easily start using. It can be used in a single machine or deployed in a cluster of machines and can be scaled horizontally as needed.
I want to try it by myself!
Trying out the MixedEmotions toolbox is really easy. Just search among our components and locate one that seems interesting to you.
For a quick test, or even for a final project that can be handled by a single machine, the only requirement is to have have Docker server installed. Once this is done, you can start downloading Docker images from MixedEmotions’ DockerHub. Then you can easily run emotion extraction services and access them via REST interface. Details are in each image description.
Sequence for using MixedEmotions’ modules from DockerHub.
If you prefer, you can also find the source of those modules from the MixedEmotions’ github repository. There are even some modules on github that do not have a Docker image yet.
In the near future you will also find info about some commercial modules for integration with the MixedEmotions platform.
I love it! I want to use it for something big!
If you try it and you like it, but you need to use this in a bigger environment, do not worry. MixedEmotions has been designed with Big Data in mind, and for that, it has been designed as a microservice architecture.
In a few words, microservice architecture are independent services. They can be replicated as needed and distributed among machines in a cluster. However, for this kind of architecture to work, there should be some components to help with the organization and communication.
To sum it up, MixedEmotions uses Marathon as a service orchestrator. Marathon launches the services, checks their status and relaunches them if they are down. This manager may relies on Mesos, a resource manager which distributes services among all the resources available in the cluster. The platform uses Mesos-DNS as a discovery service that informs of where each service is deployed.
The platform also includes an example orchestrator for processing data, one which can interact with microservices deployed in the platform and also with other external REST services.
But wait, there’s more!
The MixedEmotions platform can process emotion extraction and other characteristics from audio, video and text. But that is far from the whole picture. It also has the capabilities to analyse and represent Linked Data. For this, MixedEmotions is shipped with some additional tools. It has modules for analysing the relationship in a Social Network or to extract a knowledge graphs. It also includes Kibi, a fork of Kibana that includes the capabilities to display and filter using Linked Data.
In this graph you can see the whole picture of the MixedEmotions’ architecture. At its core there are the Docker images deployed using Marathon and Mesos. Besides emotion extraction components you can also see some of the extra services, such as the data crawlers or the Graph Data Analysis modules.
Do not worry if it seems a little bit daunting at first. We really worked towards a friendly toolbox in which each user can start slowly by just picking the components they need and then hopefully adopt more components as they need them.
If you have questions or want to get in contact, please write an email (info@mixedemotions-project.eu).
MixedEmotions is an European Research project an innovative two-year research program. It involves five companies and four European universities. With a budget of more than 3.5 million euros, it aims to search, identify, classify and characterise emotions in large volumes and data sources by applying Big Data analysis technologies.
