Project Proposal: Big Data Modeling for Data Stream Applications

This project proposal addresses the challenges of modeling big data in data stream applications, where traditional methods are insufficient due to the vast and unpredictable nature of data. The introduction highlights the importance of effective data modeling for organizing and storing massive datasets, ensuring improved performance, reduced costs, and enhanced output quality. The research background emphasizes the growth of data from internet usage and smart devices, and the limitations of traditional data modeling approaches. The research identifies the problems of unpredictable growth rates and the need for elastic data interfaces. The proposed solution introduces a feature-based modeling approach, detailing the main features of big data applications, including infrastructure, data storage, services, preprocessing, and processing/interface groups. The conclusion reinforces the effectiveness of the feature model-based approach. References to relevant literature are included, and the proposal aims to enhance understanding of data modeling techniques for students and researchers.

BIG DATA MODELLING, PROJECT PROPOSAL 1
BIG DATA MODELLING, PROJECT PROPOSAL
Student Name
Institution Affiliation
Facilitator
Course
Date

BIG DATA MODELLING, PROJECT PROPOSAL 2
Table of Contents
1.0 Introduction................................................................................................................................3
2.0 Research background.................................................................................................................3
3.0 Research problems.....................................................................................................................4
4.0 Research objectives...................................................................................................................5
5.0 Proposed modelling technique...................................................................................................5
6.0 Conclusion.................................................................................................................................7
7.0 References..................................................................................................................................7

BIG DATA MODELLING, PROJECT PROPOSAL 3
1.0 Introduction
Just like the case for other real-life applications in the library where books need to be
classified and arranged systematically on shelves so that they can be accessed easily, the massive
data handled in big data applications need an effective way to be kept in order. This process of
sorting and storing big data is called “data modelling”. In its basic definition, a data model is a
method used to organize and store data just like it is the case with the Dewey Decimal System
when it comes to organizing books in a library. Effective Data modelling in big data applications
will help arrange data in accordance with its services and usage (De Martino et al, 2014, p.650).
Through that, improved performance can be realized in big data applications considering the fact
that data models will optimize the speed of querying and reduce the input-output throughput. The
cost of handling big data will also be minimized in consideration to the fact that effective data
models will significantly reduce data redundancy. Lastly, effective data modelling approaches
will improve the quality of big data application outputs through consistent statistics and reduced
computing errors (LaValle et al, 2011, p.21).
2.0 Research background
The expansive growth in internet usage capacity, smart devices and all the other forms of
information technology in the current digital era has led to an equally impressive data growth
rate. The trend has appeared with its own challenges considering the fact that traditional
approaches to data classification, organization and storage had been designed to accommodate
small amounts of data (Lee, 2005, p.950). For instance, statistics have shown that the huge
amounts of data handled by the data stream applications cannot be stored and fully managed in a
system and that raises the concern on looking for better ways to handle data in such an
application. Additionally, contrary to traditional data, the arrival rate of the data stream has been

BIG DATA MODELLING, PROJECT PROPOSAL 4
unpredictable and thus the traditional data modelling techniques have not been in a position to
handle. For that matter, most of the traditional data modelling process does not suit or fit big data
applications due to the large amounts of data handled by these applications not forgetting the fact
that these applications deal with real-time data (LaValle et al, 2011, p.20).
3.0 Research problems
Traditional data records were characterized by their stability and predictability in their
growth and that made it relatively easy to be modelled using techniques such as UML. In
contrast, the current big data's exponential rate of growth has proved to be unpredictable because
of its myriad sources and forms. Therefore, for big data to be modelled, efforts will have to be
made in coming up with elastic and open data interfaces (Viceconti, Hunter and Hose, 2015,
p.1210). This is because changes in data sources and forms are not easily predictable. So,
considering the fact that the stability and predictability factors had not been taken care of in the
traditional data modelling techniques, most of the available data modelling approaches have been
rendered unsuitable for big data modelling.
Also, in traditional data realm, relational database schemas covered most of relationships
and data links required by business organizations for information support. This has changed with
the big data technology which sometimes doesn’t have databases or uses NoSQL databases
which don’t require database schemas (Lee, 2005, p.948). Big data models will, therefore,
require system components that have entire corporate governance and security, business
information requirements, physical storage for data, integration and open interfaces for all data
types, and have the ability to handle different types of data. The traditional data modelling
techniques have proved incapable of handling these types of functionalities.

BIG DATA MODELLING, PROJECT PROPOSAL 5
4.0 Research objectives
The aim of this research is to come up with an effective data modelling technique to be
used in modelling. This is in consideration to the fact that the traditional data modelling
approaches have proved to be ineffective because big data applications handle huge amounts of
data that cannot be stored entirely in a system and the data rates are also unpredictable in the
manner it arrives in those applications.
5.0 Proposed modelling technique
Putting into consideration the fact that big data applications handle huge amounts of data
that cannot be stored entirely in a system and the rate unpredictable rates in which data arrives in
those applications, this research proposes a different modelling approach to big data applications
rather than the normal database applications used in traditional data realm (De Martino et al,
2014, p.646). Precisely, the research has proposed a feature-based modelling approach as shown
in the diagram below

BIG DATA MODELLING, PROJECT PROPOSAL 6
(Lee, 2005, p.900)
As shown in the above figure, the main features of big data applications have been
depicted and since it’s a generic model representing a grained functionality, constraint
relationships have not been described as all the abstract features can be expanded further. On the
infrastructure side, data network group will be relied over the cable network and optionally on
wireless bases in order to support distant environmental data sensor collectors (LaValle et al,
2011, p.25). Data storage on the other side will rely on the network storages which can be
expanded to cloud storage and network clusters and optionally to the directly attached storage
mediums for temporary collection purposes.
On the side of services, the collecting group will be responsible for collecting data from
the infrastructure side either by streaming in real time and batching or static collection. The
storing group will include the logical features that depend on hardware like the database

BIG DATA MODELLING, PROJECT PROPOSAL 7
structures and storing techniques. The preprocessing group will entail data transfer, extraction,
compression, distribution, and transformation and validation operations. It is the group that will
be equipped with the most critical features for the purposes of validating the input data before
being analyzed (LaValle et al, 2011, p.22)
Lastly, the processing and interface groups will be responsible for data analysis. Where
knowledge will be derived from the raw data before visualization can follow in the interface
group. It is at these two stages that knowledge will be derived through data analytics so that it
can be represented to the end user in an understandable manner and that the end users will have a
chance to interact with the platform.
6.0 Conclusion
In this paper, the feature model-based approach has been proposed as an effective
approach to modelling big data-based applications. This is after the observation that the
traditional approaches to data modelling won’t be effective because unlike the traditional records
of data whose growth rate were predictable, big data applications handle huge amounts of data
that cannot be stored entirely in a system and the rate of data growth is also unpredictable.
7.0 References
De Martino, T., Falcidieno, B., Giannini, F., Hassinger, S. and Ovtcharova, J., 1994. Feature-
based modelling by integrating design and recognition approaches. Computer-Aided
Design, 26(8), pp.646-653.
LaValle, S., Lesser, E., Shockley, R., Hopkins, M.S. and Kruschwitz, N., 2011. Big data,
analytics and the path from insights to value. MIT sloan management review, 52(2), p.21.

BIG DATA MODELLING, PROJECT PROPOSAL 8
Lee, S.H., 2005. A CAD–CAE integration approach using feature-based multi-resolution and
multi-abstraction modelling techniques. Computer-Aided Design, 37(9), pp.941-955.
Viceconti, M., Hunter, P.J. and Hose, R.D., 2015. Big data, big knowledge: big data for
personalized healthcare. IEEE J. Biomedical and Health Informatics, 19(4), pp.1209-1215.