Social Media Optimization: Increasing the Network Size of Facebook

This document discusses the problem of increasing the network size of Facebook through social media optimization. It explores different optimization techniques such as the random forest algorithm, genetic algorithm, and expert system. The preferred approach is the expert system, which combines the advantages of the random forest and generic algorithms.

1
Quantitative Management Practice
Social media optimisation
Student’s Name
Course
Professor’s Name
Institution’s Name
Institution’s Location
Date

2
Social media optimisation: Increasing the network size of Facebook
Introduction
The media industry has been on the verge of growth and expansion with every platform
striving to have the highest subscribers. The campaign to win the number of users has been
observed among different stakeholders who utilise the social platforms for marketing,
advertising, client education on the use of the products and services, and basing on the customer
reviews for improvement (Qiao and Zhang, 2017, pp. 105-123). However, for the media
platforms to have as many subscribers as possible, several factors are considered as determinants
for growth that is measured based on the number of users. Therefore, expansion of the social
media networks is a paramount problem the firms and the media companies have been struggling
to solve to win the market shares in the industry through maximisation of the available
opportunities that would harness social media platform usage.
(Zhang and Gupta, 2018, pp.914-925) Depicted that optimisation of the social media
platform is an effective strategy for growing the network, in essence, having more subscribers
that are pertinent to market shares. Concisely, many users, for instance, on Twitter, Facebook, or
integral reflects heavy usage of the platform which in turn generates a lot of revenue to the
companies. Although optimisation has been proposed as one of the methods for solving the
challenges of growing social, media networks. Some scholars have disputed while narrowing
down the fact that the type of the optimisation technique is the cornerstone for addressing the
effectiveness in social media network growth. Therefore, the report would focus on the problem

3
of identification of optimal strategy that can be deployed to increase the network size of
Facebook (Ballings, Van den Poel, and Bogaert, 2016, pp.15-25).
Social media optimisation
(Rossmann, and Young, 2016, pp.1-52.) Defined social media optimisation as an
essential catalyst for growing a company’s online presence that entails strategic creation,
building, and maximising the social media plan for connecting to the target audience. The author
acknowledged that the growth of the social presence underlines the practices such as
strengthening the brand and generating of leads which results in getting more visible online thus
connecting to the audience.
Overview of the problem
Facebook has been experiencing an increment of the users’ way above the estimated
increase among the subscribers with approximately 1.35 billion active users monthly while 864
million active daily users (Ballings, Van den Poel, and Bogaert, 2016, pp.15-25.). This has
resulted in limitations of the Network size due to the millions of pieces of information that are
posted on the daily basis that cause the traffic overload on the network which could likely slow
down the speed of access to information. For effective retrieval and sharing of information, the
network infrastructure requires to be sufficient to enable a series of activities that to be
performed with the speeds needed by the end-users such as streaming, posting information,
downloading, and sending of all type of files to other users. A limitation of the network size
prevents the performance of the mentioned activities, which could inconvenience the users such
as the advertising and marketing companies prompting them to seek other platforms (Hamad and

4
Al-Shboul, 2017, pp. 107-117). With the emergence of the new technologies that have facilitated
fast internet speeds, the problem of small size network can barely be explained to the
subscribers, who by chance have no idea of the technical operations that are performed before
the social platform is availed. Therefore, the company has to keep up with the optimisation
techniques to increase the network size for accommodating the ever-growing number of
subscribers to retain its users as well as increasing the market shares in the media industry
(Sanni, Leemoon, Arora, and Edmonds, 2018, pp. 17-33). The summary is shown below.
Table 1: Summary of the Optimization Problem
Input Variables
User parameters for example
 Recency
 Demographic/gender
 Educational/ professional
 Geographical location etc.
Decision Variables User account type i.e. business account, user account etc.
Objective
To increase the Network size of Facebook which will
enable accommodation active millions of subscribers
accessing the platform o daily basis at high speeds.
Constraints
 Number of active users
 Relationship/ links between users
 Number of parameters
 Number of hours
Collection of data
Before conducting the mathematical computations using various models, data was
collected for the Facebook users using web crawling mechanisms where the variables such as
recency update variables (last update), demographic variables, in essence, gender, educational/
professional variables, and the geographical location variables were collected for evaluation

5
(Ballings, Van den Poel, and Bogaert, 2016, pp.15-25). The data for 5488 Facebook users were
collected representing the number of subscribers for a custom-based Facebook application with
approximately 426 variables. The summary is shown below.
Table 2: Summary of Type of Data and Source of Collection
Type of data Data Source Description
Input variable
Demographic/gender Custom-based
Facebook application
Data was collected
from the Facebook
users custom-based
applications for
individual accounts
using web crawling
mechanisms
Recency
Educational/
professional
Geographical
location
Decision variable
Business account Custom-based
Facebook applicationUser account
Solution
The literature review of available solutions
With the variety of models that researchers have deployed in various optimisation
problems, there is the need to evaluate some techniques that have been used through the
literature review to establish critics that could enable this research to settle to the most suitable
method. Several mathematical optimisation techniques are illustrated in the subsequent sections.
Random forest Algorithm

6
The random forest model has been recommended by scholars due to its ability to use the
machine-learning algorithm, which produces a great result even without the hyper-parameter
tuning (Naghibi, Ahmadi, and Daneshi, 2017, pp.2761-2775). The model is widely used due to
its simplicity and flexibility thus making it possible to be used for classification and regression
tasks. The model creates a forest through random sampling of the data variables into many
decision trees then merge them into an accurate and stable prediction as shown in the diagram
below.
Figure 1: The decision trees of Random Forest Algorithm
The model can be deployed in the evaluation of the Facebook user data by breaking the
variables into decision trees to identify the trend in the data for predictions and classification of
the variables into more stables tree decisions. Other fields that can deploy the use of the random
forest technique are the stock markets, banking, and e-commerce among others. However, as
mentioned, the primary advantage of the method is the ability to perform both classification and
regression tasks as well as its ability to produce proper prediction techniques due to its default

7
hyperparameters (Bei et al., 2018, pp.1470-1483). Contrary, the method is limited to a large
number of trees, which is likely to make the algorithm slower and less useful for real-time
predictions. The algorithms are fast to train but slow in making predictions.
Generic algorithm
The model is a metaheuristic process of natural selection whose concept belongs to the
larger class of complex evolutionary algorithms (Yu, Li, and Cai, 2016, pp.62-71). The
technique is used in generating high-quality solutions to search and optimisations problems
through relying on bio-inspired aspects such as selection, crossover, and mutation. Genetic
algorithms have been applied to populations of randomly generated individuals through iterative
processes with items in every generation referred to as a generation. Therefore, the model
identifies the fitness of every detail in the sample is evaluated whereby the eligibility is
considered as an objective function.
Furthermore, the fit items are selected stochastically from the sample whereby the
genome for every individual is modified to form a new generation, which is used in the next
iteration algorithm while terminating the undesirable characteristics. The process is repeated
until the desired items are selected. The algorithm supports multi-objective optimisation and can
work with both the discrete and continuous variables (Kovnatsky, Glashoff, and Bronstein, 2016,
pp. 680-696). The primary disadvantage is that the method is computationally expensive and
time-consuming.
Expert system

8
The expert system uses a combination of both the random forest and the genetic
algorithm (Ballings, Van den Poel, and Bogaert, 2016, pp.15-25). The technique starts by
obtaining the data from authorised access or sources, which is transformed for the identification
and selection of variables. The random forest algorithm is then used to approximate a predictive
model for the network size that in turn is utilized to evaluate the significance of specified
variable in the data sets. However, the top drivers from the predictive engine are selected as
variables, which are then optimised by the generic algorithm. In the selection, the number of
variables is determined by the parameters set by the subscribers. Additionally, during the
optimisation process, a predictive model is included in an objective function by the generic
algorithm, which is tasked to produce an optimal strategy. To increase the network size, the
administrator runs the given approach that in turn puts the optimal values as the top drivers of the
subscriber parameters (Ballings, Van den Poel, and Bogaert, 2016, pp.15-25)
Figure 2: Expert system of optimisation
Chosen approach

9
The preferred approach for increasing the Facebook network would be using the expert
system, which integrates both the random forest and the generic algorithms in solving the
optimisation problem. Through the ability to combine the approaches, the expert system is on
better position for increasing the size of the network for the users since it combines the
advantages of the two techniques as well as that of the predictive model of a network size
(Ghasab, Khamis, Mohammad, and Fariman, 2015, pp.2361-2370).
Conclusion
Increasing the size of the network is a central role for the network administrators for the
optimum utilisation of a social platform by the subscribers (Ballings, Van den Poel, and Bogaert,
2016, pp.15-25). The failure to provide the adequate size of a network could result in severe
implications in terms of the market shares due to slow speeds associated with limitation of small
size networks. The use of appropriate optimisation approach such as the expert system is highly
recommended due to its capability of combining the random forest and the generic algorithms in
increasing the size of a network through predictive and prescriptive engines into an optimal
strategy that is run by the administrator thus expanding the network size.

10
References
Ballings, M., Van den Poel, D. and Bogaert, M., 2016. Social media optimization: Identifying an
optimal strategy for increasing network size on Facebook. Omega, 59, pp.15-25.
Bei, Z., Yu, Z., Zhang, H., Xiong, W., Xu, C., Eeckhout, L. and Feng, S., 2016. RFHOC: A
random-forest approach to auto-tuning Hadoop's configuration. IEEE Transactions on Parallel
and Distributed Systems, 27(5), pp.1470-1483.
Ghasab, M.A.J., Khamis, S., Mohammad, F. and Fariman, H.J., 2015. Feature decision-making
ant colony optimization system for an automated recognition of plant species. Expert Systems
with Applications, 42(5), pp.2361-2370.
Hamad, F. and Al-Shboul, B., 2017. Exploiting Social Media and Tagging for Social Book
Search: Simple Query Methods for Retrieval Optimization. In Social Media Shaping e-
Publishing and Academia (pp. 107-117). Springer, Cham.

11
Kovnatsky, A., Glashoff, K. and Bronstein, M.M., 2016, October. MADMM: a generic
algorithm for non-smooth optimization on manifolds. In European Conference on Computer
Vision (pp. 680-696). Springer, Cham.
Naghibi, S.A., Ahmadi, K. and Daneshi, A., 2017. Application of support vector machine,
random forest, and genetic algorithm optimized random forest models in groundwater potential
mapping. Water Resources Management, 31(9), pp.2761-2775.
Qiao, R. and Zhang, X., 2017. Evolution and Development of Business Models in New Media
Environment. In New Media and China's Social Development (pp. 105-123). Springer,
Singapore.
Rossmann, D. and Young, S.W., 2016. Social media optimization: principles for building and
engaging community. Library Technology Reports, 52(8), pp.1-52.
Sanni, S.A., Leemoon, B., Arora, A.S. and Edmonds, J.J., 2018. Social Commerce Optimization:
An Integrated Framework for Consumer Behavior in Social Media. In Global Business Value
Innovations (pp. 17-33). Palgrave Pivot, Cham.
Yu, K., Li, Y. and Cai, Z., 2016. A hybrid generic algorithm for dynamic data mining in
investment decision making. International Journal on Data Science and Technology, 2(6),
pp.62-71.
Zhang, Z. and Gupta, B.B., 2018. Social media security and trustworthiness: overview and new
direction. Future Generation Computer Systems, 86, pp.914-925.