Roma +39 0644254074 - Milano +39 0240031289


Funding Scheme CP – Collaborative project
Call for Proposal
Partner Country DE AU BG SP FR HU IN IT FR PL UK
Duration 2011-2014

The recent growth in social media and user-generated content – such as Blog, Twitter, Facebook – requires high information interpretation skills, which need to be processed efficiently and appropriately. Social media and users, from a scientific point of view, pose many challenges. This is as the type of conversation is increasingly dynamic, colloquial, quick and short. In addition, the content of these news feeds are written in different languages, and the challenge is to translate them into a common one and correlate them.

Such conversations generate a stream of information that is deeply rooted in the social context and preexisting technological languages ​​and are therefore no longer adequate because they are inaccurate, scalable and portable.

The aim of the TrendMiner project is to develop an innovative and open-source methodology that allows a real-time synthesis of a multilingual information stream that is generated on social media.

Eurokleis’ task will be to adapt the generic algorithm to the financial and the economic domain and also to measure the correlation between market movements and existing views on the financial instrument.

Through the support of organisations, international caliber experts and the specific skills acquired over time concerning the development and implementation of quantitative models based on artificial intelligence and agent models, Eurokleis has developed a web application that summarises users’ views regarding the selected financial instrument. This is done in order to get an indication of the probable path of change in the short term.

This goal will be achieved through an interdisciplinary approach that combines natural language processing methods, machine learning, economics and politics. A fundamental novelty will be the use of learning algorithms for the automatic discovery of new tendencies and weakly supervised correlations. Cloud computing infrastructure will ensure accessibility and scalability.

The validity of the results will be evaluated through two high-level study cases: the support for financial decisions (through analysts, traders, regulators and economists) and the analysis and monitoring of political opinions (through politicians, economists and political journalists ). The techniques used for both of the case studies can also be used in many application areas, such as business intelligence and customer relationship management.