Search

TheOpinionMining methods are widely used to analyse and classify the choices, preferences and behaviours of consumers through the opinions gathered on the Web. On social media like TripAdvisor such opinions are usually expressed with a score and a short text. This paper proposes a strategy for ranking reviews using a scale based jointly on the rating and on the text of the reviews.

Document Clustering is the peculiar application of cluster analysis methods on huge documentary databases. Document Clustering aims at organizing a large quantity of unlabelled documents into a smaller number of meaningful and coherent clusters, similar in content. One of the main unsolved problems in clustering literature is the lack of a reliable methodology to evaluate results, although a wide variety of validation measures has been proposed. If those measures are often unsatisfactory when dealing with numerical databases, they definitely underperform in Document Clustering. This paper proposes a new validation measure. After introducing the most common approaches to Document Clustering, our attention is focused on Spherical K-means, do to its strict connection with the Vector Space Model, typical of Information Retrieval. Since Spherical K-means adopts a cosine-based similarity measure, we propose a validation measure based on the same criterion. The new measure effectiveness is shown in the frame of a comparative study, by involving 13 different corpora (usually used in literature for comparing different proposals) and 15 validation measures.

The ambiguity of words is a crucial question when dealing with an automatic analysis of documentary data bases. In Text Mining, Word Sense Disambiguation is the task of giving a particular sense to a term with different meanings both in the case of coincidental and polysemous homographs. In literature, the proposed solutions are mainly based on two elements: some knowledge related to the term and the context in which the term appears. Limiting the related knowledge to grammatical tagging, and to an analysis of collocations, here we focus our attention on identifying the context, in a data driven approach. Our framework is based on Textual Data Analysis and we assume that language and knowledge can be modeled as networks of words and the relations between them. The aim of this paper is to propose an extension of the strategy for building lexical sources in Balbi et al. (2012), in order to deal with ambiguous words. The methodological basis is given by the joint use of lexical Correspondence Analysis and Network Analysis. Our idea is investigating the neighborhood of ambiguous terms, with respect to the different latent semantic components, emerging thanks to Correspondence Analysis of a training set of documents, in order to build some rules, useful in solving WSD problems in the entire corpus.

Assuming that language can be modelled as a network of words, it is difficult to mine knowledge in textual data bases, due to their high dimensionality and the ambiguity which characterises words and their use. From a methodological viewpoint, here we propose a strategy for stressing the differences in the manifest relations emerging by Network Analysis (NA) and the latent relations obtained by lexical Correspondence Analysis (CA). Aim of this paper is to deal with the word-sense disambiguation problem, not in the usual pre-processing step, but during the analysis. The results applied to the analysis of a management commentary are presented in order to propose some statistical lexical sources, useful in the peculiar domain of business information.

Data accumulate and there is a growing need of automated systems for partitioning data into groups, in order to describe, organize and retrieve information. Dealing with documental databases, one of the main aims is text categorization, consisting in identifying documents with similar topics. In the usual Vector space mode, thel documents are represented as points in the high dimensional space spanned by words. Obstacle to an efficient performance of algorithms is the curse of dimensionality: while dimensions increase, the space where individuals are represented becomes sparse. Classical statistical methods lack their properties and interest is devoted to find dense areas in lower dimensional spaces. Aim of the paper is reviewing the basic literature on the topic, focusing attention on dimensionality reduction and double clustering.

This paper has been developed in the frame of the European project BLUE-ETS (Economic and Trade Statistics), in the work-package devoted to propose new tools for collecting and analysing data. In order to obtain business information by documentary repositories, here we refer to documents produced with non statistical aims. The use of secondary sources, typical of data and text mining, is an opportunity not sufficiently explored by National Statistical Institutes. NSIs aim at collecting and representing information in a usable and easy-readable way. The use of textual data has been still viewed as too problematic, because of the complexity and the expensiveness of the pre-processing procedures and often for the lack of suitable analytical tools. Our aim is to identify statistical linguistic sources by a deep analysis of one management commentary. From a methodological viewpoint, here we propose a tool for exploring relations between words at a micro-data level, derived from network data analysis, namely ego networks, applied together with lexical correspondence analysis.

Methods for high-dimensional data clustering represents a prolific research area in data mining, encouraging a large quantity of provisional solutions. In text mining and in the analysis of gene expression data, the idea of bidimensional clustering arose, in the sense of finding clusters of documents characterized by cluster of terms (and analogously, clusters of genes and clusters of different experimental conditions). Although we are often more interested in clustering one way of our data structure, however co clustering seems to be convenient (both from an interpretative and a computational viewpoint). Here we try to frame the problem in a multidimensional data analysis perspective, both referring to classic association and/or prediction indexes for contingency tables. Following previous works, we propose the use of a predictability index, Goodman&Kruskal tb, dealing with documents-by-terms tables. After a quick review of the wide literature related to two-way clustering, mainly developed in microarray analysis, we propose a new algorithm belonging to the genetic family, based on the optimization of the predictability index tau-b. We present experimental results to show the effectiveness of our co-clustering algorithm in practice.

Obiettivo del presente lavoro e definire delle metodologie di valutazione dell’ effi- ` cacia di un corso di laurea che consentano di identificare le aree critiche e attivare leve di miglioramento. Nello specifico l’indagine pilota si propone di delineare possibili scenari di miglioramento per il CLAS, Corso di LAurea in Statistica e informatica per l’economia e le imprese, dell’Universita Federico II di Napoli. ` Si e scelto di studiare il principale effetto di un corso universitario “efficace”: ` l’inserimento nel mondo del lavoro. Approfondendo le differenti rilevazioni sugli sbocchi occupazionali dei laureati (ISTAT, AlmaLaurea, Stella), preliminarmente e a fini di confronto e di contestualizzazione, si e cercato di dare risposta all’interrogativo di base: cosa si intende ` per “occupato”? Come viene calcolata la percentuale dei laureati occupati?

In questo lavoro si affronta il problema della comunicazione fra domanda e offerta di competenze e di professionalità in un nel mercato del lavoro particolare, quello del terzo settore. L’attenzione si è concentrata sull’erogazione di servizi legati all’immigrazione. Si è confrontato il linguaggio utilizzato dagli operatori per descri-vere, nel corso di interviste in profondità, le competenze richieste ai laureati, con il linguaggio utilizzato dalle università, per descrivere, nelle schede ministeriali, i con-tenuti e gli sbocchi professionali dei propri corsi di laurea. Da un punto di vista me-todologico il problema affrontato è quello del confronto fra corpora. Nel lavoro ven-gono presentati i principali risultati, in termini di misure di dissimilarità e rappresen-tazioni grafiche, ottenute ricorrendo a strumenti propri dell’analisi di dati simbolici e dell’analisi statistica multivariata classica.

Il lavoro si pone l'obiettivo di individuare quali sono le competenze offerte dai programmi di studio che hanno come fine la preparazione alle professioni proprie del terzo settore, attraverso uno studio statistico del linguaggio utilizzato per descrivere gli obiettivi dei corsi di laurea, sul sito del Ministero dell'Università. Il problema viene affrontato in un'ottica di text mining, vista la mole della base documentaria da analizzare e in particolare a tecniche di riduzione dimensionale che consentano l'introduzione di informazione esterna, al fine di introdurre le informazioni relative alle differente classi di laurea coinvolte e a tradizionali classificazioni delle competenze. I risultati sollevano notevoli dubbi circa la capacità delle università a porsi all'interno di un meccanismo concorrenziale

The evaluation of Public Intervention is currently considered a strategic element of political and administrative actions. The first step for evaluating public policies consists in applying the Regulatory Impact Analysis (RIA). Usually, an economical and/or a juridical viewpoint is adopted. The paper shows how quantitative methods can be useful for an ex-ante evaluation of public actions having a significant impact on people lives. A statistical approach based on Conjoint Analysis has been proposed by Scepi, et al. (2008). Thanks to this technique, developed in the frame of market research, it is possible to compare different potential actions, described by several indicators, and to estimate utilities for groups of judges. It is well-known that one of the main problems in performing a Conjoint Analysis consists in data collection. The most common procedure is the so called full profile method, which forces to deal with a small number of categorical variables (and with a small number of categories). Being based on an experimental design, it is often too rigid. Moreover, in the frame of RIA, the choice of descriptors is difficult. Here we propose a procedure based on Conjoint Analysis with textual information (Balbi, et al., 2008), where judges are asked to enrich their own choice with a free description of the desired action. An empirical analysis on alternative University systems in Italy shows the effectiveness of our proposal.

In market research, we often need statistical methods able to jointly deal with numerical and textual variables. In market segmentation, several tools have been proposed for profiling customers’ behaviours and choices, surveying a sample of potential buyers. The so called full profile method characteristic of Conjoint Analysis sometimes seems too rigid in its experimental structure. Here we propose a procedure aimed at integrating Conjoint Analysis, with textual information achieved by answers to open-ended questions. This procedure improves the understanding of the preference structure and the ideal product definition. An empirical analysis on watch market shows the evidence of this proposal.

We present a Textuale Analysis of Simbolic Objects. We apply Procrustes Rotation.

This paper aims at proposing the joint use of Canonical Correlation Analysis and Procrustes Rotations (RCA), when we deal with a text and its translation into another language. The basic idea is representing words in the two different natural languages on a common reference space. The main characteristic of this space is to be lan-guage independent, although Procrustes Rotation is performed transforming the lexical table derived from trans-lation by minimizing its distance from the lexical table belonging to the original corpus, while the subsequent Canonical Correlation Analysis treats symmetrically the two word sets. The most interesting RCA feature is building a unique reference space for representing the correlation structure in the data, inducing the two systems of canonical factors to lie on the same space. These graphical representations enables us to read distances be-tween corresponding points in terms of different way of translating the same word in relation with the general context defined by the canonical variates. Trying to understand the distances between matched points could rep-resent an useful tool for enriching lexical resources in a translation procedure. In this paper we propose the com-parison of the most frequent content bearing words in the two languages, analyzing one year (2003) of Le Monde Diplomatique and its Italian edition.

Many strategies of Text retrieval are based on Latent Semantic Indexing and its variations, by considering different weighting systems for words and documents, Correspondence Analysis and LSI share the basic algebraic tool, i.e. Singular Value decomposition and its generalisations, related to the use of different way for measuring the importance of each element, both in determining and representing similarities between documents and words. Aim of the paper is to propose a peculiar factorial approach for better visualising the relations between textual data and documents, compared with classical Correspondence Analysis. Here we consider a TF/IDF index scheme, mainly developed for Text Retrieval, in a textual data analysis context.

The paper goes through some tools considered nowadays classical in Text Mining procedures and software. We are speaking of Latent Semantic Indexing for dimensionality reduction, and the wide literature devoted to the problem of how to weight the word importance, and how to measure similarities between words and between words and queries. Visualisation is strongly affected by these choices. Here we compare some alternatives from a statistical viewpoint. A corpus consisting of six years of the Italian edition of Le Monde Diplomatique is analysed in order to show the effects of the different weighting systems together with the potentiality of Textual Data Analysis in summarising and representing newspaper information.

Costruendo regole di associazione in cui l'insieme dei candidati antecedenti e l'insieme dei candidati conseguenze sono disgiunti, consente di tenere conto di relazioni casuali fra i due insiemi. Nel lavoro, le caratteristiche delle aziende che cercano lavoro sono considerate gli antecedenti delle caratteristiche dei soggeti richiesti a fini di assunzione. Un approccio del genere all'nterno di un problema di tezxt mining degli annunci di lavoro online, consente di fare emergere le caratterstiche richieste agli attuali neolaureati

Aim of the paper is to propose a Text Mining strategy based on statistical tools, which make more efficient the extraction of information buried in massive quantities of documents. Usually, in Text Mining procedures (such as in textual data analyses) we deal with a corpus consisting of a set of documents. In order to build the data structure to be processed, each document is encoded in a document vector, according to the bag-of-words model, which associates words and their frequencies for the given document. Documents are considered as a whole. The proposed mining strategy identifies interesting sentences in the corpus we deal with, where to concentrate the knowledge extraction. The sentence interest will depend on the researcher’s objective. The proposed procedure is useful when we are interested in local contexts for words. Prior information, i.e. expert knowledge, is included, as an input for the procedure, but differently to content analysis, the key-word system is automatically built. The strategy can be applied in any case we can introduce information for partitioning documents in lower order grammatical units (e.g. sentences, but also paragraphs, etc.). The mining procedure consists in two steps: first of all the Text Categorisation, i.e. the recognition of the interesting sentences, by means of a statistical segmentation procedure, and then the knowledge extraction from the identified sub-texts. The procedure first step produces association rules useful in filtering e-mail, chat, or Web access, too. The paper aims at contributing to the day-by-day wider literature on Text Mining, devoted to go beyond the "bag-of-words" model of structuring the data set in document vectors, enhancing the role of a statistical perspective. An application on Italian on-line job offers ends the paper, showing the effectiveness of the proposal.

Questo lavoro propone una strategia per affrontare l'analisi di corpora di grandi dimensioni che presentino una strutturazione interna, con il duplice vantaggio, da un lato, di un notevole risparmio del peso computazionale dell'analisi e, dall'altro, di introdurre elementi relativi al contesto in cui le singole parole sono utilizzate. La strategia si avvale di un modesto intervento preliminare da parte di esperti al fine di pervenire, grazie all'utilizzo di metodi statistici di segmentazione, alla categorizzazione del testo e alla costruzione, sulla base di un training set, di regole di associazione, Queste regole, applicate all'intero corpus, consentono si sottoporre ad analisi soltanto i frammenti individuati come di interesse diretto per gli obiettivi conoscitivi perseguiti

This paper aiming to underline problems connected with defining and measuring Electronic Commerce. In particular, a brief review on existing statistical indicators with the problems marking them, is proposed.

The paper aims at showing the advantages of formulating lexical structures with variable elements (LSVE) in terms of symbolic objects (OS). The proposal main consequence is the collapse of the huge information usually present in a text into matrices of complex data. Thus, it is possible to use statistical tools in order to analyse the underlying properties of lexical structures by taking into account their different component distributions. Describing lexical structures with variable elements as symbolic data improves the effort of text mining by putting in a strict relation the knowledge extraction and the statistical analysis steps. The study of such lexical structures is performed by applying a factorial analysis on complex data. An application deals with a very large corpus extracted from Italian newspaper “La Repubblica” during the Nineties. In this study, we analyse the relations among the different components of some lexical structures with variable elements and the temporal evolution of some peculiar semantic traits, related to geographical-historical-political contexts.

Collana "METODI E TECNICHE PER LE RILEVAZIONI ASSISTITE DA COMPUTER"

Web 2.0 allows people to express and share their opinions about products and services they buy/use. These opinions can be expressed in various ways: numbers, texts, emoticons, pictures, videos, audios, and so on. There has been great interest in the strategies for extracting, organising and analysing this kind of information. In a social media mining framework, in particular, the use of textual data has been explored in depth and still represents a challenge. On a rating and review website, user satisfaction can be detected both from a rating scale and from the written text. However, in common practice, there is a lack of algorithms able to combine judgments provided with both comments and scores. In this paper we propose a strategy to jointly measure the user evaluations obtained from the two systems. Text polarity is detected with a sentiment-based approach, and then combined with the associated rating score. The new rating scale has a finer granularity. Moreover, also enables the reviews to be ranked. We show the effectiveness of our proposal by analysing a set of reviews about the Uffizi Gallery in Florence (Italy) published on TripAdvisor.