Search

Information management in healthcare is nowadays experiencing a great revolution. After the impressive progress in digitizing medical data by private organizations, also the federal government and other public stakeholders have also started to make use of healthcare data for data analysis purposes in order to extract actionable knowledge. In this paper, we propose an architecture for supporting interoperability in healthcare systems by exploiting Big Data techniques. In particular, we describe a proposal based on big data techniques to implement a nationwide system able to improve EHR data access efficiency and reduce costs.

Proceedings of NFMCP 2015 Workshop

Big data paradigm is currently the leading paradigm for data production and management. As a matter of fact, new information are generated at high rates in specialized fields (e.g., cybersecurity scenario). This may cause that the events being studied occur at rates that are too fast to be effectively analyzed in real time. For example, in order to detect possible security threats, researchers must screen millions of records in a high-speed flow stream. To ameliorate this problem, a viable solution is the use of data compression techniques for reducing the amount of data to be investigated. Indeed, the problem of summarizing multi-dimensional data into lossy synopses supporting the estimation of aggregate range queries has been deeply investigated in literature and many summarization approaches have been proposed, such as histograms, wavelets and sampling. In this paper, the use of summarization is investigated in a more specific context, where privacy issues are taken into account. In particular, we study the problem of constructing (in real time) \emph{privacy-preserving synopses}, that is, synopses preventing sensitive information from being extracted while supporting `safe' analysis tasks. In this regard, we introduce a probabilistic framework enabling the evaluation of the quality of the estimates which can be obtained by a user owning the summary data. Based on this framework, we devise a technique for constructing histogram-based synopses of multi-dimensional data which provide as much accurate as possible answers for a given workload of `safe' queries, while preventing high-quality estimates of sensitive information from being extracted. Moreover, we propose an efficient maintenance strategy in order to keep continuously updated histograms that can be profitably leveraged in big data scenario. To this end, we describe in this paper our system that has been used in a real life scenario.

Advances of high throughput technologies have yielded the possibility to investigate human cells of healthy and morbid ones at different levels. Consequently, this has made possible the discovery of new biological and biomedical data and the proliferation of a large number of databases. In this paper, we describe the IS-BioBank (Integrated Semantic Biological Data Bank) proposal. It consists of the realization of a framework for enabling the interoperability among different biological data sources and for ultimately supporting expert users in the complex process of extraction, navigation and visualization of the precious knowledge hidden in such a huge quantity of data. In this framework, a key role has been played by the Connectivity Map, a databank which relates diseases, physiological processes, and the action of drugs. The system will be used in a pilot study on the Multiple Myeloma (MM).

In this paper we present a study of transport properties modification of Single-Wall Carbon Nanotubes (SWNT) interacting with the most common used alkali metals. We report resistivity measurements of SWNT mats as a function of Na, K, Cs and Li doping. Our results show that as alkali exposure increases, the doped sample resistance decreases, denoting a progressive sample metallization. Furthermore, the resistivity depends upon the deposited alkali species. We hypothesize that both atomic radius and specific conductivity of doping species can affect the transport properties.

The continuous exchange of information by means of the popular email service has raised the problem of managing the huge amounts of messages received from users in an effective and efficient way. We deal with the problem of email classification by conceiving suitable strategies for: (1) organizing messages into homogeneous groups, (2) redirecting further incoming messages according to an initial organization, and (3) building reliable descriptions of the message groups discovered. We propose a unified framework for handling and classifying email messages. In our framework, messages sharing similar features are clustered in a folder organization. Clustering and pattern discovery techniques for mining structured and unstructured information from email messages are the basis of an overall process of folder creation/maintenance and email redirection. Pattern discovery is also exploited for generating suitable cluster descriptions that play a leading role in cluster updating. Experimental evaluation performed on several personal mailboxes shows the effectiveness of our approach.

Due to the emerging Big Data applications traditional data management techniques result inadequate in many real life scenarios. In particular, OLAP techniques require substantial changes in order to offer useful analysis due to huge amount of data to be analyzed and their velocity and variety. In this paper, we describe an approach for dynamic Big Data searching that based on data collected by a suitable storage system, enrich data in order to guide users through data exploration in a efficient and effective way.

The increasing availability of large process log repositories calls for efficient solutions for their analysis. In this regard, a novel specialized compression technique for process logs is proposed, that builds a synopsis supporting a fast estimation of aggregate queries, which are of crucial importance in exploratory and high-level analysis tasks. The synopsis is constructed by progressively merging the original log-tuples, which represent single activity executions within the process instances, into aggregate tuples, summarizing sets of activity executions. The compression strategy is guided by a heuristic aiming at limiting the loss of information caused by summarization, while guaranteeing that no information is lost on the set of activities performed within the process instances and on the order among their executions. The selection conditions in an aggregate query are specified in terms of a graph pattern, that allows precedence relationships over activity executions to be expressed, along with conditions on their starting times, durations, and executors. The efficacy of the compression technique, in terms of capability of reducing the size of the log and of accuracy of the estimates retrieved from the synopsis, has been experimentally validated.

Third International Workshop, NFMCP 2014, Held in Conjunction with ECML-PKDD 2014, Nancy, France, September 19, 2014, Revised Selected Papers

Due to the emerging Big Data paradigm traditional data management techniques result inadequate in many real life scenarios. In particular, OLAP techniques require substantial changes in order to offer useful analysis due to huge amount of data to be analyzed and their velocity and variety. In this paper, we describe an approach for dynamic Big Data searching that based on data collected by a suitable storage system, enrich data in order to guide users through data exploration in a efficient and effective way.

Predicting the output power of renewable energy production plants distributed on a wide territory is a really valuable goal, both for marketing and energy management purposes. Vi-POC project aims at designing and implementing a prototype which is able to achieve this goal. Due to the heterogeneity and the high volume of data, it is necessary to exploit suitable Big Data analysis techniques in order to perform a quick and secure access to data that cannot be obtained with traditional approaches for data management. In this paper, we describe Vi-POC (Virtual Power Operating Center) a distributed system for storing huge amounts of data, gathered from energy production plants and weather prediction services.We use HBase over Hadoop framework on a cluster of commodity servers in order to provide a system that can be used as a basis for running machine learning algorithms. Indeed, we perform one-day ahead forecast of PV energy production based on Artificial Neural Networks in two learning settings, that is, structured and non-structured output prediction. Preliminary experimental results confirm the validity of the approach, also when compared with a baseline approach.

The recent advances in genomic technologies and the availability of large-scale microarray datasets call for the development of advanced data analysis techniques, such as data mining and statistical analysis to cite a few. Among the mining techniques proposed so far, cluster analysis has become a standard method for the analysis of microarray expression data. It can be used both for initial screening of patients and for extraction of disease molecular signatures. Moreover, clustering can be profitably exploited to characterize genes of unknown function and uncover patterns that can be interpreted as indications of the status of cellular processes. Finally, clustering biological data would be useful not only for exploring the data but also for discovering implicit links between the objects. To this end, several clustering approaches have been proposed in order to obtain a good trade-off between accuracy and efficiency of the clustering process. In particular, great attention has been devoted to hierarchical clustering algorithms for their accuracy in unsupervised identification and stratification of groups of similar genes or patients, while, partition based approaches are exploited when fast computations are required. Indeed, it is well known that no existing clustering algorithm completely satisfies both accuracy and efficiency requirements, thus a good clustering algorithm has to be evaluated with respect to some external criteria that are independent from the metric being used to compute clusters. In this paper, we propose a clustering algorithm called M-CLUBS (for Microarray data CLustering Using Binary Splitting) exhibiting higher accuracy than the hierarchical ones proposed so far while allowing a faster computation with respect to partition based approaches. Indeed, M-CLUBS is faster and more accurate than other algorithms, including k-means and its recently proposed refinements, as we will show in the experimental section. The algorithm consists of a divisive phase and an agglomerative phase; during these two phases, the samples are repartitioned using a least quadratic distance criterion possessing unique analytical properties that we exploit to achieve a very fast computation. M-CLUBS derives good clusters without requiring input from users, and it is robust and impervious to noise, while providing better speed and accuracy than methods, such as BIRCH, that are endowed with the same critical properties. Due to the structural feature of microarray data (they are represented as arrays of numeric values), M-CLUBS is suitable for analyzing them since it is designed to perform well for Euclidean distances. In order to stronger the obtained results we interpreted the obtained clusters by a domain expert and the evaluation by quality measures specifically tailored for biological validity assessment. © 2013 Elsevier Inc. All rights reserved.

The problem of accurately predicting the energy production from renewable sources has recently received an increasing attention from both the industrial and the research communities. It presents several challenges, such as facing with the rate data are provided by sensors, the heterogeneity of the data collected, power plants efficiency, as well as uncontrollable factors, such as weather conditions and user consumption profiles. In this paper we describe Vi-POC (Virtual Power Operating Center), a project conceived to assist energy producers and decision makers in the energy market. In this paper we present the Vi-POC project and how we face with challenges posed by the specific application. The solutions we propose have roots both in big data management and in stream data mining.

Nowadays, almost all kind of electronic devices leave traces of their movements (e.g. smartphone, GPS devices and so on). Thus, the huge number of this "tiny" data sources leads to the generation of massive data streams of geo-referenced data. As a matter of fact, the effective analysis of such amounts of data is challenging, since the possibility to extract useful information from this peculiar kind of data is crucial in many application scenarios such as vehicle traffic management, hand-off in cellular networks, supply chain management. Moreover, spatial data streams management poses new challenges both for their proper definition and acquisition, thus making the overall process harder than for classical point data. In particular, we are interested in solving the problem of effective trajectory data streams clustering, that revealed really intriguing as we deal with sequential data that have to be properly managed due to their ordering. We propose a framework that allow data pre-elaboration in order to make the mining step more effective. As for every data mining tool, the experimental evaluation is crucial, thus we performed several tests on real world datasets that confirmed the efficiency and effectiveness of the proposed approach. © 2013 Springer Science+Business Media New York.

Predicting the output power of renewable energy production plants distributed on a wide territory is a really valuable goal, both for marketing and energy management purposes. Vi-POC (Virtual Power Operating Center) project aims at designing and implementing a prototype which is able to achieve this goal. Due to the heterogeneity and the high volume of data, it is necessary to exploit suitable Big Data analysis techniques in order to perform a quick and secure access to data that cannot be obtained with traditional approaches for data management. In this paper, we describe Vi-POC -- a distributed system for storing huge amounts of data, gathered from energy production plants and weather prediction services. We use HBase over Hadoop framework on a cluster of commodity servers in order to provide a system that can be used as a basis for running machine learning algorithms. Indeed, we perform one-day ahead forecast of PV energy production based on Artificial Neural Networks in two learning settings, that is, structured and non-structured output prediction. Preliminary experimental results confirm the validity of the approach, also when compared with a baseline approach.

The recent advances in genomic technologies and the availability of large-scale datasets call for the development of advanced data analysis techniques, such as data mining and statistical analysis to cite a few. A main goal in understanding cell mechanisms is to explain the relationship among genes and related molecular processes through the combined use of technological platforms and bioinformatics analysis. High throughput platforms, such as microarrays, enable the investigation of the whole genome in a single experiment. Among the mining techniques proposed so far, cluster analysis has become a standard method for the analysis of microarray expression data. It can be used both for initial screening of patients and for extraction of disease molecular signatures. Moreover, clustering can be profitably exploited to characterize genes of unknown function and uncover patterns that can be interpreted as indications of the status of cellular processes. Finally, clustering biological data would be useful not only for exploring the data but also for discovering implicit links between the objects. Indeed, a key feature that lacks in many proposed approach is the biological interpretation of the obtained results. In this paper, we will discuss such an issue by analysing the results obtained by several clustering algorithms w.r.t. their biological relevance. © 2014 Springer International Publishing.

Revised Selected Papers From NFMCP 2012

Trajectory data streams are huge amounts of data pertaining to time and position of moving objects. They are continuously generated by different sources exploiting a wide variety of technologies (e.g., RFID tags, GPS, GSM networks). Mining such amount of data is a challenging problem, since the possibility to extract useful information from this peculiar kind of data is crucial in many application scenarios such as vehicle traffic management, hand-off in cellular networks, supply chain management. Moreover, spatial data streams pose interesting challenges for their proper representation, thus making the mining process harder than for classical point data. In this paper, we address the problem of trajectory data streams clustering, that revealed really intriguing as we deal with a kind of data (trajectories) for which the order of elements is relevant. We propose a complete framework starting from data preparation task that allows us to make the mining step quite effective. Since the validation of data mining approaches has to be experimental we performed several tests on real world datasets that confirmed the efficiency and effectiveness of the proposed technique. © 2013 Springer-Verlag.

In this paper, we study the problem of mining for frequent trajectories, which is crucial in many application scenarios, such as vehicle traffic management, hand-off in cellular networks, supply chain management. We approach this problem as that of mining for frequent sequential patterns. Our approach consists of a partitioning strategy for incoming streams of trajectories in order to reduce the trajectory size and represent trajectories as strings. We mine frequent trajectories using a sliding windows approach combined with a counting algorithm that allows us to promptly update the frequency of patterns. In order to make counting really efficient, we represent frequent trajectories by prime numbers, whereby the Chinese reminder theorem can then be used to expedite the computation. © 2014 Springer International Publishing.

In this paper, we study the problem of mining for frequent trajectories, which is crucial in many application scenarios, such as vehicle traffic management, hand-off in cellular networks, supply chain management. We approach this problem as that of mining for frequent sequential patterns. Our approach consists of a partitioning strategy for incoming streams of trajectories in order to reduce the trajectory size and represent trajectories as strings. We mine frequent trajectories using a sliding windows approach combined with a counting algorithm that allows us to promptly update the frequency of patterns. In order to make counting really efficient, we represent frequent trajectories by prime numbers, whereby the Chinese reminder theorem can then be used to expedite the computation. © 2013 ACM.

RFID-based systems for object tracking and supply chain management have been emerging since the RFID technology proved effective in monitoring movements of objects. The monitoring activity typically results in huge numbers of readings, thus making the problem of efficiently retrieving aggregate information from the collected data a challenging issue. In fact, tackling this problem is of crucial importance, as fast answers to aggregate queries are often mandatory to support the decision making process. In this regard, a compression technique for RFID data is proposed, and used as the core of a system supporting the efficient estimation of aggregate queries. Specifically, this technique aims at constructing a lossy synopsis of the data over which aggregate queries can be estimated, without accessing the original data. Owing to the lossy nature of the compression, query estimates are approximate, and are returned along with intervals that are guaranteed to contain the exact query answers. The effectiveness of the proposed approach has been experimentally validated, showing a remarkable trade-off between the efficiency and the accuracy of the query estimation.

A simple hierarchical clustering algorithm called CLUBS (for CLustering Using Binary Splitting) is proposed. CLUBS is faster and more accurate than existing algorithms, including k-means and its recently proposed refinements. The algorithm consists of a divisive phase and an agglomerative phase; during these two phases, the samples are repartitioned using a least quadratic distance criterion possessing unique analytical properties that we exploit to achieve a very fast computation. CLUBS derives good clusters without requiring input from users, and it is robust and impervious to noise, while providing better speed and accuracy than methods, such as BIRCH, that are endowed with the same critical properties. © Springer-Verlag 2013.

Over the last decade, the advances in the high-throughput omic technologies have given the possibility to profile tumor cells at different levels, fostering the discovery of new biological data and the proliferation of a large number of bio-technological databases. In this paper we describe a framework for enabling the interoperability among different biological data sources and for ultimately supporting expert users in the complex process of extraction, navigation and visualization of the precious knowledge hidden in such a huge quantity of data. The system will be used in a pilot study on the Multiple Myeloma (MM). © 2012 Springer-Verlag.

XML (eXtensible Markup Language) became in recent years the new standard for data representation and exchange on the WWW. This has resulted in a great need for data cleaning techniques in order to identify outlying data. In this paper, we present a technique for outlier detection that singles out anomalies with respect to a relevant group of objects. We exploit a suitable encoding of XML documents that are encoded as signals of fixed frequency that can be transformed using Fourier Transforms. Outliers are identified by simply looking at the signal spectra. The results show the effectiveness of our approach.

The PDF format represents the de facto standard for print-oriented documents. In this paper, we address the problem of wrapping PDF documents, which raises new challenges in several contexts of text data management. Our proposal is based on a novel bottom-up hierarchical wrapping approach that exploits fuzzy logic to handle the uncertainty which is intrinsic to the structure and presentation of PDF documents. A PDF wrapper is defined by specifying a set of group type definitions that impose a target structure to groups of tokens containing the required information. Constraints on token groupings are formulated as fuzzy conditions, which are defined on spatial and content predicates of tokens. We define a formal semantics for PDF wrappers and propose an algorithm for wrapper evaluation working in polynomial time with respect to the size of a PDF document. The proposed approach has been implemented in a wrapper generation system that offers visual capabilities to assist the designer in specifying and evaluating a PDF wrapper. Experimental results have shown good accuracy and applicability of our system to PDF documents of various domains. © 2006 IEEE.

Trajectory data refer to time and position of moving objects generated by different sources using a wide variety of technologies (e.g., RFID tags, GPS, GSM networks). Mining such amounts of data is challenging, since the possibility to extract useful information from these peculiar kind of data is crucial in many application scenarios such as vehicle traffic management, hand-off in cellular networks and supply chain management. In this paper, we address the problem of trajectory data streams clustering, that revealed really challenging as we deal with data (trajectories) for which the order of elements is relevant. We propose a complete framework starting from data preparation task that allows us to make the mining step quite effective. Since the validation of data mining approaches has to be experimental we performed several tests on real world datasets that confirmed the efficiency and effectiveness of the proposed techniques. © 2011 Springer-Verlag.

The recent developments in technologies and life sciences have paved the way to complex interactions among entities in distributed and heterogeneous environments. As a result, an enormous amount of valuable information is available, spanning from structured to multimedia and spatial or spatio-temporal data. The data mining research community has been focusing on developing specialized techniques and algorithms for the analysis of such data. Some of the problems that have received attention include the analysis of multi-table data, XML data, web data, time series and sequences, graphs and trees. The first International Workshop on New Frontiers in Mining Complex Patterns (NFMCP 2011), held in Bristol UK, on September 24th 2012 in conjunction with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD 2012), was aimed at bringing together researchers and practitioners of data mining who are interested in the advances and latest developments in area of extracting patterns from complex data.

Datastreams are potentially infinite sources of data that flow continuously while monitoring a physical phenomenon, like temperature levels or other kind of human activities, such as clickstreams, telephone call records, and so on. Radio Frequency Identification (RFID) technology has lead in recent years the generation of huge streams of data. Moreover, RFID based systems allow the effective management of items tagged by RFID tags, especially for supply chain management or objects tracking. In this paper we introduce SMART (Simple Monitoring enterprise Activities by RFID Tags) a system based on outlier template definition for detecting anomalies in RFID streams. We describe SMART features and its application on a real life scenario that shows the effectiveness of the proposed method for effective enterprise management.

Radio Frequency Identification (RFID) applications are emerging as key components in object tracking and supply chain management systems. In the next future almost every major retailer will use RFID systems to track the shipment of products from suppliers to warehouses. In addition to providing insight into shipment and other supply chain process efficiencies, such data can be also valuable for determining product seasonality and other trends resulting in key information for the companies plans. Moreover, companies are already exploring more advanced uses for RFID. For example, tire manufacturers plan to embed RFID chips in tires to determine the tire deterioration. Many pharmaceutical companies are embedding RFID chips in drug containers to better track and avert the theft of highly controlled drugs. Airlines are considering RFID-enabling key onboard parts and supplies to optimize aircraft maintenance and airport gate preparation turnaround time. Due to the streaming nature of RFID readings, large amounts of data are generated by these devices. In particular, RFID applications will generate a lot of so-called “thin” data, i.e. data pertaining to time and location. This phenomenon will be even more relevant when RFIDs are so cheap that every individual item can be tagged thus leaving a “trail” of data as it moves across different locations. This scenario raises new challenges in effectively and efficiently exploiting such large amounts of data. In this paper we define a technique for detecting anomalous data in order to prevent problems related to inefficient shipments or fraudulent actions. To this end, we introduce a framework enabling users to control correct shipment of items. As an anomalous situation is detected, the system signals it in order to quickly recover the possible error. Moreover, while the trajectory is monitored we provide a validation step in order to take into account the natural “concept drift”, i.e. deviation from the shipping plan due to modification of the service. In such a case, the signalled trajectory becomes new knowledge about the overall system. In order to check whether a trajectory has to be considered anomalous with respect to past ones, we adopt a distance-based approach. Specifically, we measure similarity between item routes by comparing their Discrete Fourier Transforms. We point out that the proposed approach is flexible and can be applied in different scenarios in order to monitor very different systems independently of their spatial extension. Preliminary experiments show the effectiveness of our approach. This chapter is organized as follows. In Section 2 the problem and its application context are described. Section 3 introduces a measure of similarity between RFID series based on the Discrete Fourier Transform, which will be used to detect anomalies in object shipping as

Sensor networks represent a leading case of data stream sources coming from real-life application scenarios. Sensors are non-reactive elements which are used to monitor real-life phenomena, such as live weather conditions, network traffic etc. They are usually organized into networks where their readings are transmitted using low level protocols. A relevant problem in dealing with data streams consists in the fact that they are intrinsically multi-level and multidimensional in nature, so that they require to be analyzed by means of a multi-level and a multi-resolution (analysis) model accordingly, like OLAP, beyond traditional solutions provided by primitive SQL-based DBMS interfaces. Despite this, a significant issue in dealing with OLAP is represented by the so-called curse of dimensionality problem, which consists in the fact that, when the number of dimensions of the target data cube increases, multidimensional data cannot be accessed and queried efficiently, due to their enormous size. Starting from this practical evidence, several data cube compression techniques have been proposed during the last years, with alternate fortune. Briefly, the main idea of these techniques consists in computing compressed representations of input data cubes in order to evaluate time-consuming OLAP queries against them, thus supobtaining approximate answers. Similarly to static data, approximate query answering techniques can be applied to streaming data, in order to improve OLAP analysis of such kind of data. Unfortunately, the data cube compression computational paradigm gets worse when OLAP aggregations are computed on top of a continuously flooding multidimensional data stream. In order to efficiently deal with the curse of dimensionality problem and achieve high efficiency in processing and querying multidimensional data streams, thus efficiently supporting OLAP analysis of such kind of data, in this chapter we propose novel compression techniques over data stream readings that are materialized for OLAP purposes. This allows us to tame the unbounded nature of streaming data, thus dealing with bounded memory issues exposed by conventional DBMS tools. Overall, in this chapter we introduce an innovative, complex technique for efficiently supporting OLAP analysis of multidimensional data streams.

In this paper we propose an architecture that exploit web pages stuctural information for the extraction of relevant information from them. In this architecture, a primary role played by a distance-based classification methodology is devised. Such a methodology is based on an efficient and effective technique for detecting structural similarities among semistructured documents, which significantly differs from standard methods based on graph-matching algorithms. The technique is based on the idea of representing the structure of a document as a time series in which each occurrence of a tag corresponds to a given impulse. By analyzing the frequencies of the corresponding Fourier transform, we can hence state the degree of similarity between documents. Experiments on real data show the effectiveness of the proposed technique.

The PDF format represents the de facto standard for print-oriented documents. In this paper we address the problem of wrapping PDF documents, which raises new challenges in the information extraction field. The proposal is based on a novel bottom-up wrapping approach to extract information tokens and integrate them into groups related according to the logical structure of a document. A PDF wrapper is defined by specifying a set of group type definitions which impose a target structure to token groups containing the required information. Due to the intrinsic uncertainty on the structure and presentation of PDF documents, we devise constraints on token groupings as fuzzy logic conditions. We define a formal semantics for PDF wrappers and propose an algorithm for wrapper evaluation working in polynomial time with respect to the size of a PDF document.

A distributed system for approximate query answering on sensor network data is proposed, where a suitable compression technique is exploited to represent data and support query answering. Each node of the system stores either detailed or summarized sensor readings. Query answers are computed by identifying the set of nodes that contain (either compressed or not) data involved in the query, and eventually partitioning the query in a set of sub-queries to be evaluated at different nodes. Queries are partitioned according to a cost model aiming at making the evaluation efficient and guaranteeing the desired degree of accuracy of query answers.

Data mining algorithms generally deal with very large data sets that do not fit in main memory. Therefore, techniques that manage huge data sets need to be developed. Any algorithm that is proposed for mining data should account for out-of-core data structures. However, most of the existing algorithms have not yet addressed this issue. In this paper we describe the implementation of an out-of-core technique for the data analysis of very large data sets with the sequential and parallel version of the clustering algorithm AutoClass. We discuss the out-of-core technique and show the performance results in terms of execution time and speed up.

Often Web users want to be notified when specific information contained in a Web page has been modified. The problem of detecting Web document changes has been deeply investigated and several systems providing notification of Web page changes are available. These systems do not provide notification of changes on specific information contained in a Web page. In this work we present a system called CDWeb that performs this task. It allows users to monitor a whole document or specific portions of it. Users can also specify what kind of changes they are interested in, such as structural changes, or semantic changes. The system provides a flexible and adaptive view of the Web: it tracks user queries and creates user profiles, in order to associate a personalized view to each user.

Because of the widespread diffusion of semistructured data in XML format, much research effort is currently devoted to support the storage and retrieval of large collections of such documents. XML documents can be compared as to their structural similarity, in order to group them into clusters so that different storage, retrieval, and processing techniques can be effectively exploited. In this scenario, an efficient and effective similarity function is the key of a successful data management process. We present an approach for detecting structural similarity between XML documents which significantly differs from standard methods based on graph-matching algorithms, and allows a significant reduction of the required computation costs. Our proposal roughly consists of linearizing the structure of each XML document, by representing it as a numerical sequence and, then, comparing such sequences through the analysis of their frequencies. First, some basic strategies for encoding a document are proposed, which can focus on diverse structural facets. Moreover, the theory of discrete Fourier transform is exploited to effectively and efficiently compare the encoded documents (i.e., signals) in the domain of frequencies. Experimental results reveal the effectiveness of the approach, also in comparison with standard methods.

In this paper we propose the combined use of different methods to improve the data analysis process. This is obtained by combining inductive and deductive techniques. Inductive techniques are used for generating hypotheses from data whereas deductive techniques are used to derive knowledge and to verify hypotheses. In order to guide users in the the analysis process, we have developed a system which integrates deductive tools, data mining tools (such as classification algorithms and features selection algorithms), visualization tools and tools for the easy manipulation of data sets. The system developed is currently used in a large project whose aim is the integration of information sources containing data concerning the socio-economic aspects of Calabria and the analysis of the integrated data. Several experiments on socio-economic indicators of Calabrian cities have shown that the combined use of different techniques improves both the comprehensibility and the accuracy of models.

The urgent need for new techniques for Big data analysis calls for a great deal of attention by both research and industry communities. Among the techniques that must be redesigned for big data analysis purposes, clustering plays a crucial role. Dealing with Big Data implies that information to be analyzed have size ranging from terabytes to petabytes of data, making the use of clustering algorithms quite challenging, due to their (relatively) high computational costs. In this paper we discuss how to tackle this problem and how to implement a clustering strategy suitable for big data and having a reasonable execution time. We focus our attention on hierarchical clustering, as this class of algorithms easily meet some constraints set by big data features, while allowing the use of the most efficient solution for data access in distributed environments.