Desarrollo de un prototipo de asistente virtual para la gestión del conocimiento de una organización.
The project focuses on the creation of a virtual assistant prototype for the management of business knowledge, created with existing technologies, which seeks to support users in their knowledge requirements by ensuring that the information presented is timely. To begin the construction of the p...
| Autor Principal: | Abata Quinchuqui, Christian Paul |
|---|---|
| Otros Autores: | Ramírez Soria, Pedro David |
| Formato: | bachelorThesis |
| Idioma: | spa |
| Publicado: |
2018
|
| Materias: | |
| Acceso en línea: |
http://dspace.ups.edu.ec/handle/123456789/15875 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: |
The project focuses on the creation of a virtual assistant prototype for the
management of business knowledge, created with existing technologies, which seeks
to support users in their knowledge requirements by ensuring that the information
presented is timely.
To begin the construction of the prototype, the first step was to propose an
architecture with which the technologies were tested and from which the viability of
its development was demonstrated adhering to XP methodology.
In the first iteration data is obtained by means of a free API, integrating Bot
Framework and Luis using Node Js as a programming language, additionally
classification algorithms are investigated that will serve to discern the most important
information and knowledge for a user.
The second iteration begins with the creation of a personalized API with simulated
data, attached to a medical scenario. Although the extraction of data is correct, there
is a limitation attached to the level of maturity of the company so it changes the way
of consuming data from an API to SQL Server, on the other hand, is SVM as an
alternative algorithm to those seen in the first iteration that supposes to be the
optimum for the classification of terms.
In the third iteration, all the components are integrated and the assistant is tested with
the learning methods proposed in the second iteration, achieving successful
communication results, extraction and learning, and finally uploading the prototype
to a web environment. |
|---|