Geraldo Francisco Donega Zafalon

Spatial Big Data mining is often hindered by high computational complexity and the intrinsic autocorrelation of georeferenced records. To address these challenges, this study proposes an architectural optimization framework for the CHSMST+ algorithm, designated as CHSMST+MR. Rather than introducing a brand-new clustering paradigm, the framework focuses on a Distributed Spatial Cardinality Reduction (DSCR) layer that aggregates redundant spatial records before the core iterative mining logic begins. By transforming raw records into a weighted key-value representation within the Apache Spark environment, the proposed approach significantly mitigates the shuffling bottleneck common in distributed systems. Experimental validation using high-density biological datasets demonstrates an average execution-time reduction of 51.36%, with performance gains reaching up to 79.96% in specific high-redundancy scenarios. The results, obtained through controlled local emulation, confirm that this architectural optimization provides a scalable, deterministic, and lossless solution for accelerating spatial clustering. This work contributes a methodological path for enhancing the performance of iterative spatial mining algorithms in environments characterized by massive data density and coordinate redundancy.

The popularization of the internet allowed the use of different technologies for storing and retrieving bilingual dictionaries. An example is the Resource Description Framework (RDF) datasets. The storage of bilingual dictionaries in this type of structure makes it possible to discover indirectly connected translations through inference. However, such datasets are not efficient in graph search operations, which are important actions for inference techniques in the discovery of translations. On the other hand, database management systems based on the graph model offer a collection of resources that facilitate the tasks that involve such operations. In this way, a group of bilingual dictionaries from an RDF dataset was adapted to a Graph Database Management System (GDMS) which made it possible to reduce and simplify the structures of these dictionaries. Subsequently, 2 translation inference techniques were implemented in which two new bilingual English-Portuguese dictionaries were automatically generated. Thus, such inference algorithms are efficient in generating new dictionaries through GDMS. The scientific contribution is the application of a new technology for data storage and retrieval in support of bilingual dictionaries, in which there are advantages together with inference techniques in the generation of bilingual dictionaries and not addressed by related papers in literature.

Pattern recognition is an important field in Bioinformatics and a well-known task is the search for Single Nucleotide Polymorphism (SNP). It is possible to search for a known SNP position and analyze it using patterns of DNA bases, called masks. Nonetheless, this process becomes computationally expensive as the amount of available genomic data increases. Thus, in this study, we have developed a parallelization scheme, based on multithread programming, to SNP analysis using masks. In our tests, we noticed that the proposed scheme improved the execution time in 98.05 times when compared with the sequential approach.

The finite element method is a very reliable and precise technique for solving partial differential equations in three-dimensional domains, with relevant applications in several areas. However, 3D simulations by FEM require computer programs for solid modeling and automatic mesh generation and there are few examples of open source codes available and dedicated to these tasks. Unfortunately, these open source codes are not usually conceived to operate together in an integrated fashion, showing distinct life cycles and different origins, which may result in contradictory specifications. In this study, a method is proposed to integrate solid modeling and automatic mesh generation with focus on open source codes and how the quality of FEM simulations can be improved by the improvement of the mesh. The method was structured in desired features for the solid construction and in integration strategies for an automatic mesh generation. The approach was tested in nontrivial domains and with known relevance for studies focused on computational electromagnetics. Meshes were generated with millions of tetrahedral elements and the results were compared to the quality values commonly discussed in literature focused on FEM. Complex geometries were meshed in a few seconds, with consistent values of aspect ratios (more than 90% of the tetrahedral elements were constructed with values at most 5) and dihedral angles (the values were bounded between 5.9 to 166.7°C, with a peak value around 90°C). Finally, in order to show the relation among highly refined meshes and quality criteria which can be explored by proposed method, the Laplace's equation was simulated by FEM in order to analyze the equipotential lines of a parallel-plate capacitor. The results show how the quality of a simulation can be improved, especially concerning the increasing number of tetrahedra in the mesh with proper aspect ratio.

Co-authorship analysis in science and technology partnerships provides a vision of cooperation patterns between individuals and organizations and is still widely used to understand and assess scientific collaboration patterns. This analysis is conducted by means of bibliometry, which is the quantitative study of scientific production. However, with the evolution of database management systems, there was a significant increase in the volume of stored data, which could difficult the analysis. In this context, the developed work presents an efficient parallel optimization of bibliometric information, in order to allow this scientific analysis in a Big Data environment. Our results show that the time taken to calculate the transitivity value using the sequential approach grows 4.08 times faster than the parallel proposed approach when the number of nodes tends to infinity; the time taken to calculate the average distance and diameter values using the sequential approach grows 5.27 times faster than the parallel proposed approach when the number of nodes tends to infinity. Also, the results found present good values of speed up and efficiency.