The transition towards Industry 4.0 has led to a significant increase in the adoption of smart manufacturing, where advanced technologies, such as Artificial Intelligence and Machine Learning, are used to optimize production processes. Quality control in manufacturing presents significant challenges, particularly in detecting non-visible defects. This paper proposes a novel approach to improve quality assurance in die-casting machines for car engine block production through thermographic image analysis. Specifically, we verify whether thermal patterns in the mold, captured immediately after the part is extracted, can serve as an indicator of internal defects in manufactured components, thereby avoiding the need for expensive and time-consuming leak tests. Our approach employs a stacking ensemble as its core. The ensemble integrates Convolutional Neural Networks and Vision Transformers, leveraging their complementary strengths for defect detection. An ensemble and threshold selection process is then carried out to identify optimal classifiers for defective and non-defective parts. Experimental results based on thermographic images from a mold used in the manufacture of 4-cylinder engine blocks demonstrate that the proposed framework can ensure the internal quality of up to 63.3% of components with high confidence. This result enables a significant reduction in reliance on leak tests, illustrating the viability of a real-time, cost-effective decision-making process that reduces bottlenecks and enhances overall manufacturing efficiency.

Abstract Predictability is key for efficient and safe air traffic management. In particular, accurately estimating time of arrival for current passenger flights may help terminal controllers to plan ahead and optimize airport operations in terms of safety and resource allocation. While traditional physics-based simulations are still widely used, they are complex to model and often fail to include many factors affecting the progress of a flight. In this paper, we propose a deep learning approach based on LSTM that leverages the 4D trajectory of the flight and weather data at the destination airport, to accurately predict estimated time of arrival. We evaluate our model on flights arriving at Adolfo Suárez-Madrid Barajas airport (Spain), in the first three quarters of 2022, achieving a mean absolute error of 2.65 min over the entire flight and reporting competitive short- and long-term predictions at different spatial and temporal horizons.

Although Big RDF management is an emerging topic in the so-called Web of Data, existing techniques disregard the dynamic nature of RDF data. These RDF archives evolve over time and need to be preserved and queried across it. This paper presents v-RDFCSA, an RDF archiving solution that extends RDFCSA (an RDF self-index) to provide version-based queries on top of compressed RDF archives. Our experiments show that v-RDFCSA reduces space requirements up to 35 - 60 times over a state-of-the-art baseline, and gets more than one order of magnitude ahead over it for query resolution.

The amount of generated RDF data has grown impressively over the last decade, promoting compression as an essential tool for storage and exchange. RDF compression techniques leverage syntactic and semantic redundancies, but structural repetitions are not always addressed effectively. This paper first shows two schema-based sources of redundancy underlying to the schema-relaxed nature of RDF. Then, we revisit the W3C HDT binary format to further compact its graph structure encoding. Our HDT++ approach reduces the original HDT Triples requirements up to 2 times for more structured datasets, and reports significant improvements even for highly semi-structured datasets like DBpedia. In general, HDT++ competes with the current state of the art for structural RDF compression, leading the comparison for three of the four analyzed datasets.

Censuses are one of the most relevant types of statistical data, allowing analyses of the population in terms of demography, economy, sociology, and culture. For fine‐grained analysis, census agencies publish census microdata that consist of a sample of individual records of the census containing detailed anonymous individual information. Working with microdata from different censuses and doing comparative studies are currently difficult tasks due to the diversity of formats and granularities. In this article, we show that novel data processing techniques can be applied to make census microdata interoperable and easy to access and combine. In fact, we demonstrate how Linked Open Data principles, a set of techniques to publish and make connections of (semi‐)structured data on the web, can be fruitfully applied to census microdata. We present a step‐by‐step process to achieve this goal and we study, in theory and practice, two real case studies: the 2001 Spanish census and a general framework for Integrated Public Use Microdata Series (IPUMS‐I).

Large bilingual parallel texts (also known as bitexts) are usually stored in a compressed form, and previous work has shown that they can be more efficiently compressed if the fact that the two texts are mutual translations is exploited. For example, a bitext can be seen as a sequence of biwords ---pairs of parallel words with a high probability of co-occurrence--- that can be used as an intermediate representation in the compression process. However, the simple biword approach described in the literature can only exploit one-to-one word alignments and cannot tackle the reordering of words. We therefore introduce a generalization of biwords which can describe multi-word expressions and reorderings. We also describe some methods for the binary compression of generalized biword sequences, and compare their performance when different schemes are applied to the extraction of the biword sequence. In addition, we show that this generalization of biwords allows for the implementation of an efficient algorithm to look on the compressed bitext for words or text segments in one of the texts and retrieve their counterpart translations in the other text ---an application usually referred to as translation spotting--- with only some minor modifications in the compression algorithm.

We introduce new compressed inverted indexes for highly repetitive document collections. They are based on run-length, Lempel-Ziv, or grammar-based compression of the differential inverted lists, instead of gap-encoding them as is the usual practice. We show that our compression methods significantly reduce the space achieved by classical compression, at the price of moderate slowdowns. Moreover, many of our methods are universal, that is, they do not need to know the versioning structure of the collection.

The study of compressed storage schemes for highly repetitive sequence collections has been recently boosted by the availability of cheaper sequencing technologies and the flood of data they promise to generate. Such a storage scheme may range from the simple goal of retrieving whole individual sequences to the more advanced one of providing fast searches in the collection. In this paper we study alternatives to implement a particularly popular index, namely, the one able of finding all the positions in the collection of substrings of fixed length ($q$-grams). We introduce two novel techniques and show they constitute practical alternatives to handle this scenario. They excel particularly in two cases: when $q$ is small (up to 6), and when the collection is extremely repetitive (less than 0.01% mutations).

Bilingual parallel corpora, also know as bitexts, convey the same information in two different languages. This implies that when modelling bitexts one can take advantage of the fact that there exists a relation between both texts; the text alignment task allow to establish such relationship. In this paper we propose different approaches that use words and biwords (pairs made of two words, each one from a different text) as representation symbolic units. The properties of these approaches are analyzed from a statistical point of view and tested as a preprocessing step to general purpose compressors. The results obtained suggest interesting conclusions concerning the use of both words and biwords. When encoded models are used as compression boosters we achieve compression ratios improving state-of-the-art compressors up to 6.5 percentage points, being up to 40% faster.

Summary form only given.The hierarchical Edge-Guided techniques (called E-Gfc) enhance the original E-G approach to support high-order text statistics. These consider the same graph-based model to represent an extended input alphabet obtained by using a variant of the Re-Pair algorithm. E-Gfc adapts the previous coding scheme to grasp the features of the bit-oriented canonical Huffman code chosen as output alphabet.

This paper studies the compressibility of RDF data sets. We show that big RDF data sets are highly compressible due to the structure of RDF graphs (power law), organization of URIs and RDF syntax verbosity. We present basic approaches to compress RDF data and test them with three well-known, real-world RDF data sets.

The word-codeword mapping technique allows words to be managed in PPM modelling when a natural language text file is being compressed. The main idea for managing words is to assign them codes in order to improve the compression. The previous work was focused on proposing several mapping adaptive algorithms and evaluating them. In this paper, we propose a semi-static word-codeword mapping method that takes advantage of by previous knowledge of some statistical data of the vocabulary. We test our idea implementing a basic prototype, dubbed mppm2, which also retains all the desirable features of a word-codeword mapping technique. The comparison with other techniques and compressors shows that our proposal is a very competitive choice for compressing natural language texts. In fact, empirical results show that our prototype achieves a very good compression for this type of documents.

This paper describes a novel approach for bilingual parallel corpora (bitexts) compression. The approach takes advantage of the fact that the two texts that form a bitext are mutual translations. First, the two texts are aligned both at the sentence and the word level. Then, word alignments are used to define biwords, that is, pairs of two words, each one from a different text, that are mutual translations. Finally, a biword-based PPM compressor is applied. The results obtained compressing the two texts of the bitext together improve the compression ratios achieved when both texts are independently compressed through a word-based PPM compressor; thus, saving storage and transmission costs.