Flexible Deployment of Social Media Analysis Tools

Former COLA project partner Aragon Regional Government has developed a new web portal where the citizens will be able to apply to all the administrative procedures that they may need (almost up to 1500 different procedures). The eAdmin and Digital Society General Direction wanted to include a recommendation system to improve citizens interaction with public services. The recommendation system will be initially based on the semantic processing of information coming from different sources.

The third source of information is going to be collected using Eccobuzz integrated with MiCADOscale framework. The information will be tweets coming from institutional Twitter accounts run by Aragon Regional Government, or from topics related to Aragon posted by any other user.

Inycom has developed an application called Eccobuzz for Social Media Data Mining offered in a SaaS basis in their Data Center. Inycom was interested in MiCADOscale because Eccobuzz experiences important unpredictable fluctuations of computing load, as it depends on active customers, searches that they have configured, and information retrieved by the crawlers (which depend a lot on the topic, day of the week….). The need to have an oversized infrastructure to cope with this scenario would be mitigated thanks to MiCADOscale in a Cloud platform.

To test it, Inycom has developed a simple interface with some functionalities for MiCADOscale, called Magician, deployed on CloudSigma. The next figure displays a chart with the number of tweets collected per day. It can be seen that due to lack of computing resources (CPU), after some days working the server crashes on the April 2nd and April 10th and needs to be restarted, that is why those days the number of tweets collected is much lower.

The aim of this project for the Aragon Regional Government (represented by SARGA) was to develop a recommendation system to improve citizens interaction with public services, collecting by means of Magician the information required from public sources from the Internet (namely Twitter).

The recommendation is based on semantic analysis of different sources of information mentioned in section 2. One of them is coming from active listening carried out on public information (Twitter and Internet) related to topics of interest for the Government of Aragon (eAdministration, tourism, employment, events…) and Govermental Twitter accounts. This will be done using Eccobuzz technology (Magician) adapted to MiCADOscale.

This recommendation system is planned to be extended in the near future to other sources of information such as the historical applicant's behaviour (citizens who did this also did that).

The main technical impact of this use case is the flexibility of the approach that MiCADOscale enables for a type of application with very unpredictable and fluctuating CPU load, minimizing the computing infrastructure requirements, specially for those customers using the application on-premise and/or in private clouds.

Also, the container based architecture facilitates the development and deployment of applications by software programmers.

Inycom intends to offer in a SaaS basis different commercial applications using MiCADOscale, not only Eccobuzz. This will bring:

Aragon Government will improve citizens life offering better services. It is a social impact, not economic, as it is a non-profit governmental organization.