The increasing prevalence of node-compromising attacks in Internet of Things (IoT) networks threatens the integrity, availability and reliability of data-driven decision-making. To address this, we propose SHIELD, an AI-driven self-healing framework that ensures robust detection and automatic mitigation of compromised nodes in large-scale IoT environments. SHIELD relies on the Isolation Forest algorithm for unsupervised anomaly detection, enabling accurate identification of malicious behavior without the need for labeled datasets. Beyond detection, SHIELD integrates a fully autonomous self-healing mechanism that revokes compromised nodes and dynamically reconfigures the network to maintain resilience and performance. Experimental evaluations demonstrate the effectiveness of SHIELD: it achieves a detection accuracy of 99.00%, surpassing traditional models such as MLP, SVM, and CNN. Notably, after an attack, SHIELD’s self-healing process significantly enhances detection performance — increasing recall from 35.29% to 82.35%, F1 score from 52.17% to 90.32%, and AUC-ROC from 67.64% to 91.17%. Moreover, SHIELD significantly enhances network efficiency by reducing overall energy consumption by 20.36%. These findings highlight SHIELD’s dual capability to deliver robust security and energy-efficient self-healing, positioning it as an ideal solution for resource-constrained IoT infrastructures.

Cancer pathology reports are integral to cancer diagnosis and treatment. The growing volume of these reports necessitates automated classification, prompting a shift towards rule-based Natural Language Processing (NLP) systems due to their capacity for more contextualized processing. This review examines these rule-based approaches, highlighting the key features extracted from the pathology reports, and evaluating the benefits they offer. Despite certain limitations, rule-based methods demonstrate considerable promise in the classification of cancer pathology reports. We conclude by identifying future research opportunities aimed at addressing these limitations to further enhance the benefits of the classification process.

The emergence of the Internet of Things (IoT) networks as a source of large amount of data has paved the way for the adoption of machine learning models. Divers datasets, used in the training phase, are issued by collected data from deployed IoT networks. This has attracted the attention of adversaries seeking to exploit these models for their gain. The adversaries compromise IoT/sensor nodes to manipulate these models through poisoning attacks wherein they introduce carefully malicious data into the model’s training dataset. In this paper, we propose a framework, namely NoComP for Node Compromising detection, to defend against poisoning attacks by detecting compromised nodes and delete their readings from the collected data. NoComP prevents datasets to be mixed poisonous collected sensed data. To this end, we use as machine learning algorithm the neural network to detect compromised nodes. This algorithm offer significant advantages in terms of efficiency and accuracy in detecting anomalies. We carry out experiments to evaluate NoComP and compare it with two existing proposals. The accuracy and efficiency results shows that NoComP outperforms the existing ones and improves the robustness against poisoning attacks.

International audience

Horizontal partitioning is an optimization technique applied to improve query processing time in distributed data warehouses. Scanning a large number of HDFS data blocks, to respond to the ad-hoc or OLAP queries, is a heavy operation. We can skip loading unnecessary data blocks if we partition or index some tables by the appropriate predicate attributes. However, the way of selecting the candidate's attributes remains a challenging task to handle. In this paper, we propose a technique based on frequent itemset mining, to Partition, Bucket and Sort the Tables (PBSTs) of a big data warehouse with the more frequent predicate attributes in the queries. We take into account the density of the attributes of the tables, data skew, and the physical characteristics of the cluster nodes. To evaluate our approach, we conducted some experiments in a cluster of 15 slave nodes. Experimental results show that with our method, we improve the query response time by 50 % over existing skipping techniques.

The amount of data that is captured and generated by modern computing devices has augmented exponentially over the last years. The Hadoop framework - an open source project based on the MapReduce paradigm - is a popular choice for processing these large volumes of data or big data. However, the performance gained from Hadoop's features is currently limited by its default block placement policy, which does not take any data characteristics into account. This is particularly true for relational data bases and data warehouses. Indeed, the efficiency of many operations can be improved by a careful data placement, including indexing, grouping, aggregation and joins. In this paper we propose a data warehouse distribution strategy to improve query gain performances on multi-nodes clusters, especially Hadoop clusters. Based on k-means clustering method that allows to master the number of clusters through its k parameter, we investigate the performance gain for OLAP cube construction with and without data organization. And this, by varying the number of clusters and data warehouse size. Our experiments suggest that a good data placement on a cluster during the implementation of the data warehouse increase significantly the OLAP cube construction and querying performances.

Parallel computing is a fundamental technique in the management of large quantities of data as it leverages on the concurrent utilization of multiple computing resources. One of the technologies that made big data analytics popular and accessible to enterprises of all sizes is MapReduce (and its open-source Hadoop implementation). With the ability to automatically parallelize the application on a cluster of commodity hardware, MapReduce allows enterprises to analyze terabytes and petabytes of data more conveniently than ever. However, the performance gained from Hadoop's features is currently limited by its default block placement policy, which does not take any data characteristics into account. Indeed, the efficiency of many operations can be improved by a careful data placement, including indexing, grouping, aggregation and joins. In this paper, we present a MapReduce data blocks allocation approach to improve MapReduce jobs execution and query performances on multi-nodes clusters, especially Hadoop clusters. Based on k-means clustering method that allows to master the number of clusters through its k parameter, we study the influence of number of clusters on queries execution instead of queries performances with and without data organization. For this, we used well-known, large-scale data analysis benchmark: TPC-H. Our experiments suggest that defining a good data placement on a cluster during the implementation of a data warehouse increase significantly the OLAP cube construction and querying performances.

Nowadays, many governments and states are involved in an opening strategy of their public data. However, the volume of these opened data is constantly increasing, and will reach in the near future limitations of current treatment and storage capacity. On the other hand, the MapReduce paradigm is one of the most used parallel programming models. With a master-slave architecture, it allows parallel processing of very large data sets. In this paper, we propose a parallel approach based on Mapreduce to process public open data. Applied, as a case study, to the official data sets from the French Ministry of Communication. We implement a parallel algorithm as a solution to define a ranking of national museums and galleries according to the accessibility degrees for people with disabilities. We studied the feasibility of our approach in two main parts: The performance in terms of execution time, and, the visualization of the obtained results in order to integrate them into solutions such as geographic BI. This work can be applied to other cases with very large data sets.

The emergence of large volumes of data imposed by the major players of the web requires new management models and new data storage architectures and treatment able to find information quickly in a large volume of data. The column-oriented NoSQL (Not Only SQL) database provide for big data the most suitable model to the data warehouse and the structure of multidimensional data in OLAP cube form. However, in the absence of OLAP cube computation operators, we propose in this paper, a new aggregation operator called CN-CUBE (Columnar NoSQL CUBE), which allows data cubes to be computed from data warehouses stored in column-oriented NoSQL database management system. We implemented the CNCUBE operator using the SQL Phoenix interface of HBase DBMS and conducted experiments on a public data warehouse in a distributed environment produced using the Hadoop platform. Thus we have shown that our CN-CUBE operator has OLAP cubes computation times very suitable for NoSQL warehouses.

This paper presents a multidimensional model and a language to construct cubes for the purpose of on-line analytical processing. Both the multidimensional model and the cube model are based on the concept of complex object which models complex entiti

Microblogging sites have become a staple in our modern world. They provide the users with the ability to keep in touch with their contacts, using up of 140 characters in the case of Twitter sites. Responding to this emerging trend, it becomes critically important to interactively view and analyze the massive amount of microblogging data from different perspectives and with multiple granularities. In the area of Business intelligence, On-line analytical processing (OLAP) is a powerful primitive for data analysis. However, OLAP tools face major challenges in manipulating unstructured text such as microblogging data.

Traditional data warehousing technologies and On-Line Analytical Processing (OLAP) are unable to analyze textual data. Moreover, as OLAP queries of a decision-maker are generally related to a context, contextual information must be taken into account during the exploitation of data warehouses. Thus, we propose a contextual text cube model denoted CXT-Cube which considers several contextual factors during the OLAP analysis in order to better consider the contextual information associated with textual data. CXT-Cube is characterized by several contextual dimensions, each one related to a contextual factor. In addition, we extend our aggregation OLAP operator for textual data ORank (OLAP-Rank) to consider all the contextual factors defined in our CXT-Cube model. To validate our model, we perform an experimental study and the preliminary results show the importance of our approach for integrating textual data into a data warehouse and improving the decision-making.

In this paper, we cross two techniques for performance tuning of an XML cube. We analyze six configurations for managing the cube. The configurations result from storing two variants of the cube (unfragmented and fragmented) in different ways. First, we consider a disk-resident database. Then, we consider caching the frequent properties of the unfragmented cube and the frequent fragments of the fragmented cube. Finally, we load and manage the entire cube into the main memory. We show the benefits of vertical fragmentation and in-memory management of the XML cube through a set of experiments.

Summary form only given. In order to prepare complex data for relevant analysis, a data warehousing-based approach is needed. However, a good multidimensional modeling requires efficient preparation of data starting with a data integration phase. We present in this paper two principal steps of the complex data warehousing process. The data integration is the first one. To do that, we define a generic UML data model capable of representing a wide range of complex data including their possible semantic properties. Furthermore, complex data are represented as XML documents generated through an implemented prototype. The second important phase is the preparation of data for the multidimensional modeling. We demonstrate that we can use data mining techniques to help the user in building a better multidimensional model.