As the presence of artificial intelligence has increased in a variety of different areas, the use of machine learning and deep learning techniques for creative purposes has also risen significantly in recent years. Works of this kind within the area of natural language processing (NLP) are typically neural models used for fiction or lyrics generation. Those works are in most cases in English and adapting them to other languages is not feasible. In this work, we develop a Spanish text generator system for the rap sub-genre known as freestyle. Freestyle songs present unique challenges for text generation given that performers compete with one another in a lyric improvisation contest. Given the low availability of freestyle text, especially in Spanish, we collected two separate datasets, one with freestyle lyrics and the other, larger, with rap lyrics, which are more readily available. The rap dataset can be used for pretraining, and the freestyle dataset for finetuning on the generation task. Furthermore, we design a neural network-based generation model that takes into account both the structure of freestyle and the low data availability. The model was able to generate realistic freestyle verses in Spanish.

There is consensus that the best way for reducing insolvency situations in financial institutions is through good risk management, which involves a good client selection process. In the market, there are methodologies for credit scoring, each analyzing a large number of microeconomic and/or macroeconomic variables selected mostly depending on the type of credit to be granted. Since these variables are heterogeneous, the review process carried out by credit analysts takes time. The objective of this article is to propose a solution for this problem by applying fuzzy logic to the creation of classification rules for credit granting. To achieve this, linguistic variables were used to help the analyst interpret the information available from the credit officer. The method proposed here combines the use of fuzzy logic with a neural network and a variable population optimization technique to obtain fuzzy classification rules. It was tested with three databases from financial entities in Ecuador — one credit and savings cooperative and two banks that grant various types of credits. To measure its performance, three benchmarks were used: accuracy, number of classification rules generated, and antecedent length. The results obtained indicate that the hybrid model that is proposed performs better than its previous versions due to the addition of fuzzy logic. At the end of the article, our conclusions are discussed and future research lines are suggested.

This paper presents an experiment to show the importance of causal sentences in summaries. Presumably, causal sentences hold relevant information and thus summaries should contain them. We perform an experiment to refute or validate this hypothesis. We have selected 28 medical documents to extract and analyze causal and conditional sentences from medical texts. Once retrieved, classic metrics are used to determine the relevance of the causal content among all the sentences in the document and, so, to evaluate if they are important enough to make a better summary. Finally, a comparison table to explore the results is showed and some conclusions are outlined.

Currently, data repositories contain a plethora of information in different formats, most of which consists of text. This situation has raised interest in the study of techniques to automate the identification of the most relevant sentences of a document with the goal of generating a text summary. This article presents a technique for extracting the most representative sentences in a document, employing a user-defined criteria. The criteria is learned by the system using an optimization technique and a training document where the user has ranked the sentences according to their relevance. The proposed method has been applied to a five-chapter thesis with good results. At the end of this paper we provide some conclusions as well as ideas for future work.

Educational Data Mining collects the various methods that allow extracting novelty and useful information from large data volumes in educational contexts. This paper describes the process used to, through Data Mining techniques, identify the most relevant characteristics in relation to student academic performance at the School of Computer Science of the National University of La Plata. The results obtained using the proposed method to process the information relating to regular and non-regular students at the UNLP allowed establishing interesting relationships in relation to student academic performance. Based on the obtained models it can be said that the fact that the student works does not mean that their academic performance decrease and young students that take several years to join the faculty have better performance if they express interest in getting a job.

Particle Swarm Optimization (PSO) is a metaheuristic that has been successfully applied to linear and non-linear optimization problems in functions with discrete and continuous domains. This paper presents a new variation of this algorithm - called oscPSO - that improves the inherent search capacity of the original (canonical) version of the PSO algorithm. This version uses a deterministic local search method whose use depends on the movement patterns of the particles in each dimension of the problem. The method proposed was assessed by means of a set of complex test functions, and the performance of this version was compared with that of the original version of the PSO algorithm. In all cases, the oscPSO variation equaled or surpassed the performance of the canonical version of the algorithm.