Integration and analysis of unstructured data towards database optimization and decision making using deep learning techniques
Loading...
Date
2024-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Kampala International University
Abstract
This thesis addresses the challenge of integrating unstructured data into a Relational Database
Management System (RDBMS). The increasing volume and variety of unstructured data pose
significant challenges for organizations seeking to leverage such data for decision-making.
Traditional RDBMS are not well-equipped to handle unstructured data due to their structured nature,
leading to inefficiencies in data storage and analysis. To overcome these challenges, a model is
developed to automatically integrate unstructured data into a Relational Database Management System
(RDBMS). The objectives include designing a classification model, implementing it for data
integration and analysis, optimizing it for database optimization and decision support, and validating
its effectiveness. The model efficiently extracts relevant information from categorized unstructured
documents, facilitating structured database construction. The study rigorously followed a data science
research methodology, encompassing data collection, model development, implementation, testing,
evaluation, and validation. Results show significant performance improvement with the incorporation
of LSTM layers, notably achieving an accuracy boost from 83.2% to 94.6% in receipt image
processing. Similar improvements were observed across precision, recall, and F1-Score metrics. This
accomplishment substantially addressed the hurdles associated with processing and analyzing
unstructured data. In conclusion, the researcher strongly recommends the adoption of this model for
the analysis of unstructured data. Future research could focus on further optimizing the model's
performance and scalability, exploring additional deep learning techniques, and extending its
applicability to other domains.
Description
A thesis submitted to the school of mathematics and computing in partial fulfilment of the requirement for the award of a degree of Master of Science in software engineering of Kampala international university