The project requires the integration and analysis of highly diverse data types: structured (GPS coordinates), semi-structured (badge logs, camera metadata), and unstructured (WebEx messages, email attachments). NoSQL (Not only SQL) databases are specifically designed for this purpose. They offer flexible schemas that can easily accommodate varied data formats without the rigid structure of traditional relational databases. This flexibility is crucial for big data and machine-learning applications that ingest data from multiple, heterogeneous sources, making NoSQL the appropriate choice.

A. RDBMS: Requires a predefined, rigid schema, making it inefficient and impractical for storing and querying unstructured data like free-form text messages and binary email attachments.

B. object-oriented: While it can handle complex data, this model is less common and generally not the primary choice for the scale and variety of data ingestion described in big data scenarios.

D. flat file: Lacks the robust querying, indexing, scalability, and concurrent access capabilities necessary for a sophisticated, large-scale machine-learning application.

1. Cisco Official Documentation: Cisco

"Cisco Data Intelligence Platform White Paper

" 2020. Page 4

under the "Ingest and Process" section

discusses the platform's architecture for handling "data in any format and from any source

including streaming

batch

and real-time." This capability to manage heterogeneous data is a hallmark of systems built upon NoSQL principles.

2. University Courseware: Abadi

D.

"CS 186: Introduction to Database Systems

Lecture 21: NoSQL

" University of California

Berkeley

Fall 2015. Slide 4

"Why NoSQL?"

explicitly lists "Flexible

schema-less data models" as a primary reason for adopting NoSQL

which directly addresses the challenge of handling the mixed data types (logs

messages

GPS) in the question.

3. Peer-Reviewed Publication: Edlich

S.

& Friedland

A.

et al. "NoSQL: A la découverte des bases de données nouvelle génération

" academie-de-lyon.fr

2011. Chapter 1

Section 1.2

"Les limites du modèle relationnel

" explains that relational models are poorly suited for applications requiring the storage of large volumes of complex

non-uniform data

such as that found in social networks or large-scale analytics projects

which is analogous to the scenario presented.