The implemented Social Business Intelligence system makes it possible to interpret and analyse unstructured texts generated and collected from the web, to extract qualitative and quantitative assessments. Specifically, the system exceeds the limitations currently featured by all Web Monitoring and Sentiment Analysis systems, namely: the inability to integrate the data flow and relevant possible analyses with the company's IT system. the non implementability of analyses simultaneously based on internal data and on unstructured data extracted from the web The Social BI system makes it possible to browse the data extracted from the web, semantically enriched downstream of a Text Mining process, highlighting what is being said, how often and above all highlighting the semantic polarisation, thus making it possible to exploit the information value of these Big Data.

Download PDF version
Matteo Golfarelli
Stefano Rizzi
Area di specializzazione
Culture and creativity
sentiment analysis
social business intelligence
opinion mining
web monitoring
Presentazione grafica degli argomenti più discussi sul web
Innovative aspects

A Social BI system makes it possible to perform OLAP analyses on unstructured or semi-unstructured data, an option not provided by any BI tool currently available on the market. Thanks to the platform, the flow of information from the web may be integrated with the company's internal proprietary data, allowing more information to be extracted, through exploitation of the informational potential encapsulated in social-origin Big Data. The demonstrator is based on the use of a semantic engine for lexico-syntactic-semantic text analysis (unstructured information).

Potential applications

In the marketing realm, a Social BI system is able to gather and analyse the opinions expressed on a product, an advertising campaign or any other company aspect, and aid in assessing the economic return produced by a certain investment - hence also aid in planning future investments. From the point of view of product development, understanding how it is perceived by one's clientèle, the criticism and most appreciated aspects provides a priceless competitive advantage

OLAP multidimensional analysis of concepts extracted from internet conversations
Application example

Implementation of a Web Monitoring and unstructured data analysis system to monitor the return on investments in marketing campaigns

Application description and results

By way of example, it was opted for developing the platform in a case study in the sphere of politics. The developed application consists of three main functional modules: a web monitoring and crawling module which implements a process, based on boolean queries using inherent keywords of the specific listening environs one wishes to analyse: a textual analysis and semantic text enrichment module an interface module consisting of interactive dashboard and reports to analyse the gathered data. The web monitoring and crawling module makes use of a service implemented by the British firm Brandwatch, thanks to which the text of potential interest for the application is recovered and stored on the application's local DB. The textual analysis and text enrichment module is supported by an external semantic engine, developed by SyNTHEMA of Pisa, on which an intense activity was carried out for adaptation to the specific application environs of the linguistic resources used. Standard technologies in web development such as Javascript, PHP and HTML5, on the other hand, have been used to implement the user interface, in order to make the analysis and user experience as interactive, appealing and intuitive as possible

Involved partners


Implementation Time
8 months (1 person) + 4 months (2 people)
Technology Readiness Level
TRL7 - System prototype demonstration in operational environment

Our demonstrator arises from the many years of experience of CIRI-ICT in the field of Data Warehousing, Business Intelligence and technological transfer and experimentation activity in applying these new tools in new emerging markets

Sentiment and topic occurrences trends over time