Furio Camillo

Monitoring the Web Sentiment, The Italian Prime Minister's Case

The world has fundamentally changed as the Internet has become a universal means of communication. The Web is a huge virtual space where to express individual opinions and influence any aspect of life. Internet contains a wealth of data that can be mined to detect valuable opinions, with implications even in the political arena. Nowadays the Web sources are more accessible and valuable than ever before, but most of the times the true valuable information is hidden in thousands of textual pages. Their transformation into information is therefore strongly linked to their automatic lexical analysis and semantic synthesis. This poster describes a Knowledge Mining study performed on over 1000 news articles or posts in forum/blogs, concerning the Italian Prime Minister Silvio Berlusconi, involved last year in the sexual scandal. All these textual contributions have been Morpho-Syntactically analysed, Semantically Role labelled and Clustered in order to find meaningful similarities, hilite possible hidden relationships and evaluate their sentiment polarity.

Mining the Web to monitor the Political Consensus

Communication is becoming more and more crucial in the competitive political arena: politicians can monitor electors’ suggestions or claims, or the perception they might have about leaders’ statements, by analyzing blogs, newsgroups and newspapers. They try to take account of the complexity of public views in order to design populist measures and increase dramatically their consensus. The Web sources are more accessible, ubiquitous, and valuable than ever before. But the most valuable information is often hidden and encoded in blog posts or pages, which are often neither structured, nor classified, being free textual. The process of accessing all these raw data, heterogeneous in terms of source and lexicon, and transforming them into information is therefore strongly linked to automatic textual analysis and conceptual synthesis. This paper describes a Sentiment Mining study performed on over 1,000 news articles or forum/blog posts, concerning the Italian Prime Minister Silvio Berlusconi, involved in the escorts’ scandal.

Subjective Business Polarization: Sentiment Analysis Meets Predictive Modeling

The growth of Internet and the information technology has generated big changes in subjects communication, that, nowadays, occurs through social media or via thematic forums. This produced a surge of information that is freely available: it offers the possibility to companies to evaluate their credibility and to monitor the ”mood” of their markets. The application of Sentiment Analysis (SA) has been proposed in order to extract, via objective rules, positive or negative opinions from (unstructured) texts. Communication literature, instead, highlights how such polarization derives from a subjective evaluations of the texts by the receivers. In business applications the receiver (i.e. marketing manager) is leaded by the values and the mission of the company. In our paper we propose a strategy to fit brand image and company values with a subjective SA, a probabilistic Kernel classifier has been employed to get discrimination rule and to rank classification results.

On the Choice of the Kernel Function in Kernel Discriminant Analysis Using Information Complexity

In this short paper we shall consider the Kernel Fisher Discriminant Analysis (KFDA) and extend the idea of Linear Discriminant, Analysis (LDA) to nonlinear feature space. We shall present a new method of choosing the optimal kernel function and its effect on the KDA classifier using information-theoretic complexity measure.

Advances in credit scoring: combining performance and interpretation in kernel discriminant analysis

Due to the recent financial turmoil, a discussion in the banking sector about how to accomplish long term success, and how to follow an exhaustive and powerful strategy in credit scoring is being raised up. Recently, the significant theoretical advances in machine learning algorithms have pushed the application of kernel-based classifiers, producing very effective results. Unfortunately, such tools have an inability to provide an explanation, or comprehensible justification, for the solutions they supply. In this paper, we propose a new strategy to model credit scoring data, which exploits, indirectly, the classification power of the kernel machines into an operative field. A reconstruction process of the kernel classifier is performed via linear regression, if all predictors are numerical, or via a general linear model, if some or all predictors are categorical. The loss of performance, due to such approximation, is balanced by better interpretability for the end user, which is able to order, understand and to rank the influence of each category of the variables set in the prediction. An Italian bank case study has been illustrated and discussed; empirical results reveal a promising performance of the introduced strategy.

Graduates’ employment and employability after the “Bologna Process” reform. Evidence from the Italian experience and methodological issues

In a phase of depression and systemic crisis investments are essential assets in organizing the recovery, and the more so when innovation is relevant. This is why universities, companies, households and graduates implement strategies for overcoming the present crisis, leading to structural changes and competition both at the local and international level. In this framework, tracer studies on graduates transition to the labour markets provides fundamental insights and information not only to the organizations responsible for their training, but also to the economic system as a whole. Moreover, any such study is all the more useful when it can draw upon reliable and up-to-date information. This paper emphasizes three main points. First we present the results achieved by the AL model in tracing the transition path of graduates from the time they enrolled at the university until a few years after earning the degree. The survey is carried out every year by the AL and makes it possible to analyze the most recent labour market trends through the scrutiny of the career opportunities available for the graduates after 1, 3 and 5 years on from graduation. More specifically, we will present the results of the 2008 survey. This survey involved also all first and second level graduates from the 2007 vintage. Second, we examine the revision in our survey method, adopted in order to face the need to monitor a much higher number of post-reform graduates (more than 140 thousand overall) and the call of the Ministry and the universities to keep the information as much detailed as possible in assessing the employment outcomes for each single degree course, without losing feasibility in terms of costs and data collection time. In fact, we resorted to a mixed method: the computer assisted web interviewing (CAWI) and the computer assisted telephone interviewing (CATI). This is why it became necessary to measure and assess the effect of this approach on the answers given by interviewed graduates. In third place, we outline the results of some preliminary experiments carried on in order to allow for specific and recurrent comparisons between the results achieved with the AL model and other similar models dealing with the employment conditions of Italian graduates.

Semiometric Approach, Qualitative Research and Text Mining Techniques for Modelling the Material Culture of Happiness

Drawing from a recent ethnographic research on Happiness carried throughout 8 European countries in the 2003/4, Future Concept Lab will illustrate how the use of interactive digital material can be relevant to analyse qualitative and quantitative data in a participatory and creative manner. Our speech will focus on the additional value of presenting data in an interactive and flexible way by using a two-ways insight matrix and a word mapping statistical technique called Semiometrie. In order to exemplify their usage, we will draw on a recent research “The Material Culture of Happiness” based on the collection and the analysis of photo diaries coming from Spain, France, England, Germany, Italy, The Netherlands, Finland and Russia.

Assessing Balance of Categorical Covariates and Measuring Local Effects in Observational Studies

This paper presents a data driven approach that enables one to obtain a global measure of imbalance and to test it in a multivariate way. The main idea is based on the general framework of Partial Dependence Analysis (Daudin, 1981 J. J.) and thus of Conditional Multiple Correspondences Analysis (Escofier, B. (1988). Analyse des correpondances multiples conditionelle. La Revue de Modulad) as tools for investigating the dependence relationship between a set of observed categorical covariates (X) and an assignment-to-treatment indicator variable (T), in order to obtain a global imbalance measure (GI) according to their dependence structure.We propose the use of suchmeasure within a strategy whose aimis to compute treatment effects by subgroups. A toy example is presented for illustrate the performance of this promising approach.

Osservatorio del mercato del lavoro della provincia di Bologna: Rapporto 2006

The Report presents the main informations about labour market in the county of Bologna during 2006. The Bologna labour market goes fairly well: in front of a positive business cycle employment is still growing and Bologna county has met many of the Lisbon targets. Nevertheless, in a comparison with European regions some weaknesses are shown, mainly for what concerns human capital formation and flexibility of firms’ organization. Moreover, the quality of employment is worsening, as a growing percentage of new jobs are on of a fixed term kind.

Kernel Discriminant Analysis and information complexity: advanced models for micro-data mining and micro-marketing solutions

In this paper we shall consider Kernel Discriminant Analysis as an innovative tool for supervised classification in a business vision as a marketing solution. The main idea we propose is the combined use of information complexity and bootstrap process which allows the user to overcome the open problems of such a technique as the kernel function choice and at the same time check the robustness of the rule found.