CHAPTER and has prompted a development from

CHAPTER ONE

Google’s Eric
Schmidt reported that we had achieved the point where more data was being made
each two days than in all of mankind’s history up to 2003 (Siegler, 2010). From
that point forward, organizations of every kind imaginable, over an extensive
variety of businesses, have been getting to grips with better approaches for
dealing with and utilizing the extraordinary volume of data that is being
produced each day.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

According to
Marr (2015), users of Facebook, the biggest social media and networking site on
the planet, transfer billions of bits of content to the social network site
each day.

The term ‘big
data’ as defined by Taylor-Sakyi (2016) refers to large sets of complex data,
both structured and unstructured which traditional processing techniques and/or
algorithms are unable to operate on. Its aim is to uncover shrouded patterns
and has prompted a development from a model-driven science paradigm into a
data-driven science paradigm.

However,
outfitting and overseeing big data isn’t just about the big players and their
multitudes of data researchers crunching through gigantic data sets. It is for
every individual as well. Along these lines, big data management is of utmost
significance to social media giants like Facebook. The employ the use of Apache
Hadoop Software to organize and manage their data.

Along these
lines, Cynthia Harvey (2017) defined Big data management as a wide idea that
incorporates the policies, procedures and technology utilized for the
collection, storage, governance, organization, administration and delivery of
large repositories of data. It can incorporate data purging, relocation,
joining and arrangement for use in reporting and analytics.

She additionally
portrayed that big data management is firmly identified with the possibility of
data lifecycle administration (DLM). This is a strategy based approach for
figuring out which data ought to be stored inside an association’s IT
environment, and additionally when data can be securely erased.

Data, regardless
of whether it is small or big or a mix of the two, matters to each and every
individual or organization in each industry. To take into account this
tremendous interest for data, many organizations like Facebook have sprung up
offering administrations that empower different organizations to dispatch data
activities and bridle the power of data without putting resources into costly
technology or hiring scientific staff. Extracting big data from Facebook also
involves the use of R-programming. Many High Level Programming languages like
Python, Ruby, PHP, etc, offer tools to enable data scientist and individuals
extract these data seamlessly.

Data extracted
from Facebook posts are peoples’ opinions with respect to issues influencing
their day by day lives. As indicated by Shapshak (2016), Nigeria’s month to
month dynamic users of Facebook have moved to 16 million (from 15 million)
which is a 6.67% expansion, as of April, 2016. Organizations, associations and
also prominent personalities and government officials are currently utilizing
Facebook for corporate branding, online activism, applauding or calling them
out and even instigating the society at large to help in instances of illness
outbreaks, protest campaigns and numerous others.

Associations,
politicians, and other important personalities utilize social media, Facebook
in particular, to create discussions with their target audience, relating to
their services, products or campaigns. If done right, this leads to an increase
in traffic, more buzz of the brand fine projection of party competitors. This
abundance of social data is more often than not in an unorganized, divided and
informal form, making it hard to get useful data from the sites as users have to
spend a considerable measure of energy and time manually filtering the data and
some of the time, end up not getting the proposed message. This issue opens a
door for enhanced data mining through NLP sentiment analysis.

In spite of the
fact that critical research endeavors have been placed in sentiment
classification and analysis, the vast majority of the current techniques depend
on natural language processing tools to parse and analyze sentences in a survey,
yet they offer poor precision, on the grounds that the writing in online reviews
have a tendency to be fragmented and less formal than writing in news or journal
articles.

Numerous opinion
sentences contain grammatical errors and unusual terms that are not found in
dictionaries. Not at all like Twitter which is constrained to 140 characters,
Facebook status messages can take well more than 60,000 characters (Cohen,
2011) allowing users to better write-ups and a more precise depiction of
feelings.

Opinion mining
or sentiment analysis is a discipline related with information mining that makes
use of machine learning methods and natural language processing to figure out
what a specific group of individuals feel about an issue. Computers can make
use of machine learning, natural language procedures and statistics to perform automated
sentiment analysis of digital messages on large collection of texts, including webpages,
online news, online surveys, web blogs, social networking and discussion groups
on the internet. This research work looks to survey one machine learning algorithm
for sentiment on the Buhari Support Organization (BSO) Facebook page.

1.2.         STATEMENT
OF THE PROBLEM

The obvious
popularity and value of social networks has prompted the formation of
tremendous quantities of printed information often in an unstructured, fragmented
and informal form.

Organizations,
governments et cetera commonly get high volumes of electronic feedback from the
public each day, some of it as evoked reviews, some as spontaneous/unsolicited
remarks, suggestions and fault-finding. The huge number of reviews make it
difficult for organizations or governmental establishments to respond to
criticism rapidly and to guide it to the appropriate channels inside the
organization/association for action.

Further, readers
or potential customers of these organizations will most likely be unable to
rapidly make educated choices in light of the unstructured data and may give up
as it might take too long to go through the information so as to identify the information
required to make a choice. To address the above difficulties, it is important to
give an intelligent and computerized system that can adequately organize and
group social media data so it can be utilized by human users, definitively.

This would
essentially give valuable information ranging from rates of consumer loyalty to
public opinion trends to policy makers which can enable them save cash and
enhance consumer satisfaction. Further, extracting the sentiment of a survey
can help give compact summaries to readers and automatically create helpful recommendations
for them.

While much work
has as of late centered around the analysis of social media keeping in mind the
end goal to discover what individuals think about current topics of interest,
there are, nonetheless, still many challenges to be confronted. Among such is
the obvious challenge in extracting a singleton sentiment label from long
passages, where fluctuations over the length of the document make
characterization difficult (Ssoriajr, Kanej 2010). Likewise, given the usual
and unstructured nature of social media data, it is important to play out an analysis
of the different machine learning approaches for sentiment analysis to see
which approach performs best or is the most useful for the sort of information
accessible on social media.

This research tackles
the issue of characterizing Facebook posts based on the mindsets communicated
in them (positive, negative or nonpartisan) and producing summaries and
patterns of the classified data. These summaries can discover relevance in
different regions e.g. helping customer care service centers in keeping better customer
relations or helping government organizations to breakdown and rapidly respond to
concerns of the public or politicians and public figures, to get a general
perspective of the supporters’ opinion and take the necessary action.

1.3.         AIMS
AND OBJECTIVES

The objectives
of this study therefore are to:

1.      
Develop a system for extracting storing Facebook
data.

2.      
Analyze machine learning algorithms utilized as
a part of other characterization models and assess their suitability for the
issue of grouping Facebook posts for sentiment analysis.

3.      
Develop classifiers by utilizing the algorithms identified
above and extract features that will enable them to classify sentiments into
the positive, negative or neutral.

1.4.         PURPOSE
OF THE STUDY

The purpose of
this study is to get more insight into the impact of big data management of
Facebook and extraction of data for analytics. Keeping in mind the end goal to
accomplish this, literatures study are conducted to examine the existing works
and the findings from previous studies.

In particular,
data from Buhari Support Organization (BSO), Nigeria’s President’s political
care group on Facebook, is collected and a NLP-based sentiment analysis is
carried out on the data.

1.5.         SIGNIFICANCE
OF THE DATA

This research reveals
that Big data is enormous, comes at a speed and exceptionally unstructured that
it doesn’t fit regular relational database structures. With so much insight
embedded in this data, an alternative method to process this massive data is
essential. Huge companies like Facebook are all around resourced to deal with
this task but the measure of data generated each day effortlessly exceeds this
limit. Less expensive equipment, cloud computing and open source technologies have
made processing of big data significantly less expensive.

In addition, in
a couple of lines of codes, and numerous accessible free assets on the web,
Python has turned out to be a more worthy web scraping programing language. Its
implementation in this research work has in fact made data extraction on
Facebook seamless.

Once more, this
study is noteworthy in light of the fact that it will include work that will
prompt having an apparatus that can help organizations, government or people to
sort out feedbacks or review data from the masses in a more valuable and meaningful
way, thereby helping in enhancing service delivery and consumer satisfaction.
All the more, create a tool that will give an analysis of the different classification
methodologies, for example, Naïve Bayes, Maximum Entropy and Support Vector
Machines to help decide the best characterization approach for Facebook/Social
media data.

1.6.         SCOPE
OF THE STUDY

The focus in
this research is on the challenges of big data management and the connection
between big data and Facebook data scraping for analytics. Along these lines,
this study adds to scientific research on the areas of management and influences
of Facebook data extraction analytics on the general public.

It gives data scientists
new bits of knowledge into how organizations manage the advantages and challenges
of big data management.

1.6.         RESEARCH
QUESTION

This research
examines Facebook big data management comprehensively and data extraction model
for analytics. The ultimate inquiry of the research is:

“Can the
activities of users on Facebook have significant impact on the result of a
process, like elections in the society?”

Go Top
x

Hi!
I'm Rita!

Would you like to get a custom essay? How about receiving a customized one?

Check it out