Vito Santarcangelo

Gender recognition: Methods, datasets and results

Digital Out Of Home (DOOH) applications which exploit computer vision algorithms to automatically collect soft biometrics of people in front a smart screen are of great interest for industry. In the last years many gender recognition pipelines have been proposed in literature. Different benchmark datasets have been introduced and used for testing purpose. This paper gives an overview of the state-of-the-art in the context of gender recognition by highlighting features, classifiers and datasets which can be employed to reach the goal. Comparisons of the results obtained by different approaches are also presented.

Social Opinion Mining: An Approach for Italian Language

Opinion Mining or Sentiment Analysis is an important Computer Science topic, that allows the discovery of the web users attitude on one or more topics. Marketing, Social Engineering and Information Retrieval are achieved through the application of Opinion Mining on Web Data. Today Social Networks represent one of the most important places on the web for sharing information, media and opinions (26% of people have accounts on Social Networks). Integration with the opinions on social networks and social communities can improve the relevancy and quality of Opinion Mining Tools. This paper discusses how Social Networks Mining represents one of the most important tasks for Opinion Mining Systems, considering the state of art literature. Then, we will show an interesting approach based on ADJECTIVES (A), INTENSIFIERS (I) and NEGATIONS (N) (called AIN) developed for Italian. This approach is based on the use of an Italian Sentiment Thesaurus (AIN Thesaurus), developed by the authors and presented in this work.

Innovative Methods for the Development of a Notoriety System

The role of internet in our society is growing day by day and is becoming more and more the only way for getting information, exchange opinions and for improving our personal culture. So, an huge mole of data, in all fields, is today easily accessible and everybody can express and exchange ideas. This represents the greatness of the web. But at the same time, to this huge amount of data does not always correspond an appropriate quality of information that we are reading, and nowadays this represents the biggest weakness of the web. Aim this the work is to analyze the approach based on marked chain used by Pagliarani et al. as we explain in the introduction, showing the relation of this method with BigData analysis.

Big Data and forensics

Nowadays, it is easy to trace a large amount of information on the web, to access documents and produce a digital storage.The current work is submitted as an introduction to an innovative system for the investigation about notoriety of web data which is based on the evaluation of judicial sentences and it is implemented to reduce the duration of all processes.This research also aims to open some new conjoint debates about the study and application of statistical and computational methods to web data on new forensics topics: text mining techniques enable us to obtain information which may be helpful to establish a statistical index in order to describe the quality and the efficiency in terms of law. It is also possible to develop an intelligent system about facts and judgments.

The Sentiment of the Infosphere: A Sentiment Analysis Approach for the Big Conversation on the Net

In the Network Society the use of hashtags has become a daily routine for the participation on the Big Conversation Iorio and Ruoto (Nessun tempo, 2015). Designated by a ‘hash’ symbol (#), a hashtag is a keyword assigned to information that describes it and aides in searching. Hashtags are now central to organize information on Social Networks. Hashtags organize discussion around specific topics or events and they are becoming an integrated part of the Infosphere, the whole informational environment constituted by all informational entities. The sentiment analysis of Hashtags shared on the Big Conversation can return a possible snapshot about the sentiment shared by users. Scope of this work is to present an application of sentiment analysis on the Italian hashtags of mainly social networks as part of the ‘Infosphere’. This analysis returns a semantic sentiment report about the hashtags shared by the users of the social networks, that can produce a semantic sentiment trend about users. This approach could be applied to every language simply changing the sentiment thesaurus used.

Market basket analysis from egocentric videos

Abstract This paper presents Visual Market Basket Analysis (VMBA), a novel application domain for egocentric vision systems. The final goal of VMBA is to infer the behavior of the customers of a store during their shopping. The analysis relies on image sequences acquired by cameras mounted on shopping carts. The inferred behaviors can be coupled with classic Market Basket Analysis information (i.e., receipts) to help retailers to improve the management of spaces and marketing strategies. To set up the challenge, we collected a new dataset of egocentric videos during real shopping sessions in a retail store. Video frames have been labeled according to a proposed hierarchy of 14 different customer behaviors from the beginning (cart picking) to the end (cart releasing) of their shopping. We benchmark different representation and classification techniques and propose a multi-modal method which exploits visual, motion and audio descriptors to perform classification with the Directed Acyclic Graph SVM learning architecture. Experiments highlight that employing multimodal representations and explicitly addressing the task in a hierarchical way is beneficial. The devised approach based on Deep Features achieves an accuracy of more than 87% over the 14 classes of the considered dataset.

Big Data Compliance for Innovative Clinical Models

Abstract In the healthcare sector, information is the most important aspect, and the human body in particular is the major source of data production: as a result, the new challenge for world healthcare is to take advantage of these huge amounts of data de-structured among themselves. In order to benefit from this advantage, technology offers a solution called Big Data Analysis that allows the management of large amounts of data of a different nature and coming from different sources of a “computerized” healthcare, as there are considerable changes made by the input of digital technology in all major health areas. Clinical intelligence consists of all the analytical methods made possible through the use of computer tools, in all the processes and disciplines of extraction and transformation of crude clinical data into significant insights, new purposes and knowledge that provide greater clinical efficacy and best health pronouncements about past performance, current operations and future events. It can therefore be stated that clinical intelligence, through patient data analysis, will become a standard operating procedure that will address all aspects of care delivery. The purpose of this paper is to present clinical intelligence approaches through Data Mining and Process Mining, showing the differences between these two methodologies applied to perform “real process” extraction to be compared with the procedures in the corporate compliance template (the so called “Model 231”) by “conformance checking”.

Egocentric Vision for Visual Market Basket Analysis

This paper introduces a new application scenario for egocentric vision: Visual Market Basket Analysis (VMBA). The main goal in the proposed application domain is the understanding of customers behaviours in retails from videos acquired with cameras mounted on shopping carts (which we call narrative carts). To properly study the problem and to set the first VMBA challenge, we introduce the VMBA15 dataset. The dataset is composed by 15 different egocentric videos acquired with narrative carts during users shopping in a retail. The frames of each video have been labelled by considering 8 possible behaviours of the carts. The considered cart’s behaviours reflect the behaviour of the customers from the beginning (cart picking) to the end (cart releasing) of their shopping in a retail. The inferred information related to the time of stops of the carts within the retail, or to the shops at cash desks could be coupled with classic Market Basket Analysis information (i.e., receipts) to help retailers in a better management of spaces and marketing strategies. To benchmark the proposed problem on the introduced dataset we have considered classic visual and audio descriptors in order to represent video frames at each instant. Classification has been performed exploiting the Directed Acyclic Graph SVM learning architecture. Experiments pointed out that an accuracy of more than 93 % can be obtained on the 8 considered classes.