Bibliographic citations
Cueva, A., Cueva, C., Huacac, J., Junco, A. (2023). Modelo de medición de desempeño de servicios financieros en Perú a través de análisis de sentimiento utilizando métodos de decisiones multicriterio [Pontificia Universidad Católica del Perú]. http://hdl.handle.net/20.500.12404/25716
Cueva, A., Cueva, C., Huacac, J., Junco, A. Modelo de medición de desempeño de servicios financieros en Perú a través de análisis de sentimiento utilizando métodos de decisiones multicriterio []. PE: Pontificia Universidad Católica del Perú; 2023. http://hdl.handle.net/20.500.12404/25716
@mastersthesis{renati/537676,
title = "Modelo de medición de desempeño de servicios financieros en Perú a través de análisis de sentimiento utilizando métodos de decisiones multicriterio",
author = "Junco Navarro, Ananda Adelaida",
publisher = "Pontificia Universidad Católica del Perú",
year = "2023"
}
This study aims to create a model for measuring the performance of financial services using Natural Language Processing (NLP) techniques like Sentiment Analysis (SA) and a Multicriteria Decision Method. (MCDM) enables the evaluation of products or services offered by Peru's four major banks by analyzing positive, negative, or neutral opinions expressed on the social network Twitter. The study identifies a set of aspects or criteria for scoring the sentiment expressed in tweets, generates a consolidated score table, and finally evaluates the performance of each banking entity based on the identified aspects. The research was established within a design framework as applied research, using a quantitative approach, since it seeks to solve a practical problem based on the theory and then provide an alternative source of information that supports a bank client's decision-making when hiring a financial service to improve satisfaction. For the model's construction, 15,546 tweets were evaluated, with only 5,276 chosen that mentioned at least one aspect. The data set was analyzed using sentiment polarization methods such as Stanza, VADER, TextBlob and BETO (BERT), and then the MCDM VIKOR provides additional information to determine the best service option to contract. The results obtained show us that the best method to determine the polarity of the tweets is BETO, achieving an accuracy level of 88%, in addition to verifying that both BETO and VIKOR meet the acceptability conditions defined by the methods; achieving this adequately resolve the best service alternative.
This item is licensed under a Creative Commons License