OSWALDO Solarte-Pabon

Dr. OSWALDO Solarte-Pabon


Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 April 2024 in Multimedia Tools and Applications
Verified
Journal Article
Open access
PDF file
ÁGÁlvaro García-Barragán
Álvaro García-Barragán
ACAlberto González Calatayud
Alberto González Calatayud
OSOSWALDO Solarte-Pabon
OSWALDO Solarte-Pabon
Mariano ProvencioMariano Provencio
Mariano Provencio
Ernestina MenasalvasErnestina Menasalvas
Ernestina Menasalvas
VRVíctor Robles
Víctor Robles
In recent years, there has been a remarkable surge in the development of Natural Language Processing (NLP) models, particularly in the realm of Named Entity Recognition (NER). Models such as BERT have demonstrated exceptional performance, leveraging annotated corpora for accurate entity identification. However, the question arises: Can newer Large Language Models (LLMs) like GPT be utilized without the need for extensive annotation, thereby enabling direct entity extraction? In this study, we explore this issue, comparing the efficacy of fine-tuning techniques with prompting methods to elucidate the potential of GPT in the identification of medical entities within Spanish electronic health records (EHR). This study utilized a dataset of Spanish EHRs related to breast cancer and implemented both a traditional NER method using BERT, and a contemporary approach that combines few shot learning and integration of external knowledge, driven by LLMs using GPT, to structure the data. The analysis involved a comprehensive pipeline that included these methods. Key performance metrics, such as precision, recall, and F-score, were used to evaluate the effectiveness of each method. This comparative approach aimed to highlight the strengths and limitations of each method in the context of structuring Spanish EHRs efficiently and accurately.The comparative analysis undertaken in this article demonstrates that both the traditional BERT-based NER method and the few-shot LLM-driven approach, augmented with external knowledge, provide comparable levels of precision in metrics such as precision, recall, and F score when applied to Spanish EHR. Contrary to expectations, the LLM-driven approach, which necessitates minimal data annotation, performs on par with BERT’s capability to discern complex medical terminologies and contextual nuances within the EHRs. The results of this study highlight a notable advance in the field of NER for Spanish EHRs, with the few shot approach driven by LLM, enhanced by external knowledge, slightly edging out the traditional BERT-based method in overall effectiveness. GPT’s superiority in F-score and its minimal reliance on extensive data annotation underscore its potential in medical data processing.

1 September 2018 in Scalable Computing: Practice and Experience
Verified
Journal Article
Closed access
DSDavid Sanchez
David Sanchez
OSOSWALDO Solarte-Pabon
OSWALDO Solarte-Pabon
VBVictor Andres Bucheli
Victor Andres Bucheli
HOHugo Ordoñez
Hugo Ordoñez
The aim of this paper is to present a method based on the isoefficiency model for assessing the scalability in big data environments. The programs word count and sort were implemented and compared in Hadoop and Spark. The results confirm that isoefficiency presented a linear growth as the size of the data sets was increased. They were checked experimentally to ensure that the evaluated frameworks are scalable and a sublinear function was obtained. This paper discusses how scalability in big data is governed by a constant of scalability (beta).

11 October 2024 in Preprints.org
Verified
Preprint
Open access
PDF file
VBVictor Andres Bucheli
Victor Andres Bucheli
CGCarlos Gaona
Carlos Gaona
OSOSWALDO Solarte-Pabon
OSWALDO Solarte-Pabon
This article examines the impact of network degree distribution on the performance of evolutionary algorithms, specifically in solving the Traveling Salesperson Problem (TSP). It evaluates the integration of various network types to regulate population crossovers, enhancing candidate solutions and their refinement over generations. The experimental analysis includes complex networks like Erdős–Rényi and Barabási–Albert, along with regular graphs such as Balanced trees and Harary graphs. The study focuses on how different network structures influence exploration, convergence, and computational efficiency, particularly in terms of CPU time. Results show that highly connected networks, like Barabási–Albert, facilitate faster convergence and improved solution quality by guiding the search process more effectively. The Harary graph with k=2 demonstrated superior performance, reducing execution times by up to sevenfold compared to traditional approaches. This improvement highlights the advantages of specific degree distributions in balancing exploration and exploitation. Networks with hubs, such as Barabási–Albert, accelerate information flow and the evolutionary process, while more regular networks, like Harary graphs, provide controlled structures that maintain high solution quality. These findings emphasize the importance of network topology in evolutionary algorithms and suggest further exploration of hybrid models and other network types to optimize NP-hard problems.