Unpacking The Black Box An Exploration of The Algorithms Driving Social Media Networks-A Review

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Unpacking the Black Box: An Exploration of the

Algorithms Driving
Social Media Networks-A Review
Amey Thorat Amulya Kura
Department of Computer Engineering Department of Computer Engineering
Vidyalankar Institute of Technology Vidyalankar Institute of Technology
Ansika Jaiswal Sapana Survase

Department of Computer Engineering Department of Computer Engineering
Vidyalankar Institute of Technology Vidyalankar Institute of Technology
Amit Aylani
Prof., Department of Computer Engineering
Vidyalankar Institute of Technology
Abstract:- Social media algorithms are used for finding professionals in your field, you may utilize social media to
detailed information in large unstructured data by advance your professional network and increase your
relevant keywords used by users. There are different expertise in a certain subject. Social media gives your
algorithms used for social media from a searching point business the opportunity to interact with customers, get their
of view. One of the algorithms is the "Probability of opinion, and build brand recognition.
Node's Degree" algorithm, which is based on the concept
of breadth-first search, random walk, and the highest Online social networks (OSNs) have recently gained
degree seeking algorithm. The algorithm involves considerable attention. For example, Twitter had over 300
selecting a source node and a target node, and then million monthly active users in 2018. There are many studies
traversing the nodes in the network to find the target that have analyzed a particular OSN as a graph with nodes of
node. The algorithm checks if the target node is a users and edges of relationships among users.
neighbor of the current node and, if not, transmits a
query message to other nodes based on their probability In this paper, we will discuss the findings of various
of being relevant to the search. Nodes with higher degrees researchers on different algorithms related to complex
are more likely to be searched, making the algorithm networks. The first paper by GU Yiran and ZHAO Wenwen
beneficial to nodes with higher degrees. In addition to introduces the concept of the Probability of Node's Degree
this, there are other algorithms such as FP-FOREST, algorithm that is based on breadth-first search, random walk,
DSTree, UPTree algorithm, and KC-LA, which are used and the highest-degree seeking algorithm. The second paper
for finding frequent patterns, maintaining and mining discusses the development of a compact data structure named
frequent item sets, and finding K-Clique in complex FP-FOREST that enhances the performance of an existing
social networks. These algorithms are useful in data- algorithm called INSTANT for frequent pattern mining in
driven decision-making and in gaining insights into social social media streams. The third paper by Mohammad Mehdi
media analytics. Daliri Khomami et al. proposes a distributed learning
automata-based algorithm called KC-LA for finding K-
Keywords:- Social Media Algorithm, Social Media Analytics, Clique in complex social networks. This paper covers various
Complex Social Network, Social Media, K-Clique, Learning algorithms proposed for solving maximal clique finding,
Automation, Betweenness Centrality, Random Walk. which is an NP-hard problem, with practical applications in
community detection in social networks. The fourth paper
I. INTRODUCTION discusses the estimation of top nodes which has highest
betweenness centrality which has a shortest path pass
Social Media has become an integral part of our daily through the vertices. In the following sections, we will
lives, allowing people to connect, share, and exchange discuss each paper and highlight their contributions to the
information and ideas in virtual communities and networks. field of complex networks.
Social media may help us connect with friends and

family, learn about and pursue new hobbies, and have fun
from a personal viewpoint. By engaging with other
IJISRT23MAY444 www.ijisrt.com 819

ISSN No:-2456-2165
II. LITERATURE SURVEY unordered data, these fundamental data structures are helpful.
A hash table, for instance, offers instant access to data that is
A. Probability Of Node’s Degree On Complex Networks indexed by any key value, which could be a dictionary, a
In this Paper, author GU Yiran, ZHAO Wenwen number (such as a memory address for cached memory), or a
proposed the social media’s search algorithm “Probability of URL. Graphs are used to represent relationships between
Node’s Degree” which is based on the concept of breadth objects, and this course covers a variety of graph
first search, random walk, highest degree seeking algorithm. representational data structures and graph traversal
algorithms, such as determining the shortest path between
Before going on an improved algorithm of probability two nodes. This course will also address disjoint sets, another
of node’s degree first we will see the based algorithm of idea on which these graph algorithms will depend.data
concept. Breadth First Search (BFS) begins as a level order structure and associated algorithms.
traversal. Basically, in this algorithm, we can find the path
between two people in level order. Random Walk algorithm C. Finding K-Clique In Complex Social Networks
is used for extracting information from the ensemble of paths The selected paper titled “Distributed Learning
between persons in a graph. Process of searching for the Automata-based Algorithm for Finding K-Clique in Complex
highest-degree neighbor in DS in each node is aware of the Social Network” authored by Mohammad Mehdi Daliri
information about its neighbors. However, it also somewhat Khomami et al. Proposes a new algorithm that utilizes
lessens the search process's query message flow, which can distributed LAs to solve k-Clique called(KC-LA)in social
help prevent network congestion. networks.
Now, based on the above concept, the probability of a The authors discuss the problem of maximal clique
node's degree algorithm is explained, which is given above: finding in graphs, which involves finding a subset of nodes in
Each node knows the degree of its use neighbors, as shown a graph where each node is connected pairwise. The
by the image of the search by degree sequence, and searches maximum clique refers to the clique with the largest number
for the neighbor with the highest degree at each step. The of nodes. This problem is NP-hard, and various algorithms
neighbor with the second-highest degree will be picked to have been proposed to solve it, including k-clique
broadcast a query message if the neighbor with the highest algorithms, which involve finding cliques with a fixed size k.
degree has already been visited. The search technique can Many of these algorithms fall into one of three categories:
thus get the best possible results when used on complex deterministic, heuristic, and approximation. Maximal clique
networks. finding has practical applications in fields such as
community detection in social networks. Some researchers
 Algorithm of probability of node's degree search (PDS) have attempted to utilize the concept of cliques for
works as follows: community detection, and various algorithms have been
This algorithm involves selecting a source node and a developed for this purpose. Additionally, the authors mention
target node, and then traversing the nodes in the network to LA-based algorithms that have been successful in solving
find the target node. The algorithm checks if the target node graph problems such as positive influence dominating set,
is a neighbor of the current node and if not, transmits a query independent set, vertex cover, and community detection.
message to other nodes based on their probability of being
relevant to the search. Nodes with higher degrees are more Now based on the above concept, a learning automaton
likely to be searched, making the algorithm beneficial to is a type of learning model that learns to choose the best
nodes with higher degrees. Nodes can be visited more than action from a set of actions in a random environment by
once, edges can be visited only once. The search continues receiving feedback in the form of rewards or penalties. It
until the target node is found or all the nodes have been updates its probability vector for selecting actions based on
visited. whether the feedback is favorable or unfavorable. A DLA is
a network of LA that cooperates to solve a particular
B. Finding Frequent Patterns In Social Media Streams problem. In this work, just a LA is deactivated at a ume. The
A cutting-edge algorithm called INSTANT is being number of actions performed by automata is equal to the
improved with the help of a small data structure called FP- number of LAS Connected to it [3].
FOREST, which demonstrates how to compress itemsets and
count supports efficiently. The algorithm performs better in  KC-LA based for k-Clique -
terms of memory use and execution time, according to the The KC-LA algorithm finds a k-clique in a social graph
results. By utilising its appealing qualities, a novel tree by using a network of learning automata. Each automaton is
structure known as DSTree (Data Stream Tree) gathers assigned to a vertex and chooses actions corresponding to the
significant data from streams and can be readily maintained edges connected to the vertex. The algorithm selects an
and mined for frequent itemsets as well as numerous other automaton, activates it, and adds it to the k-clique set. The
patterns such restricted automaton selects an action, and the algorithm checks
whether it can be added to the k-clique set. The process
 UPTree algorithm - continues until all automata are disabled. The algorithm
Hash tables, disjoint sets, and graphs can all be evaluates the cardinality of the k-clique set and rewards or
implemented using the data structures and algorithms penalizes the selected action based on whether it increases or
covered in the Unordered Data Structures course. For decreases the size of the k-clique set. The algorithm also

ISSN No:-2456-2165
modifies the probability of selecting actions based on the method which has an overall high collection ratio. In the
degree of nodes to improve the selection process. second stage the top-k nodes are estimated from a gram
which has highest betweenness centrality. In this paper the
D. Estimation of Top-K Betweenness Centrality Nodes author proposed a new crawling based algorithm which first
In the paper “Estimating Top-k Betweenness Centrality estimates the ego betweenness centrality by random walk and
Nodes in Online Social Networks”, author proposed a new then estimates the top-k betweenness centrality.
crawling-based algorithm to estimate the top-k nodes in
online social networks which has the highest betweenness The betweenness centrality is the sum of the ratio of
centrality. Privacy and security concerns prevent obtaining shortest paths between two nodes which pass through a node
complete information in Online social networks hence in the graph. The users or nodes which have high
Crawling based algorithm is used which considers limitation betweenness centrality in online social networks are the main
of request. This algorithm works in two stages: the first is a key for the spread of information. Since these nodes have
sampling method in which edges or nodes are sampled from high betweenness centrality, hence they are present on many
a graph. They describe that they should choose a sampling of the shortest paths in the graph.
III. COMPARISON
Probability of Node’s Finding Frequent Finding K- Clique

Estimation of Top-k
Degree Patterns between ness centrality
TOPIC nodes
WHY IS IT NEED ED? To Send Message using Mining associations , Community detection, Finding the nodes which
highest degree node correlations, relationships solving graph problems plays bridge role in
among data network
LIMITATIONS Only works for Mining frequent patterns Identifying largest Ignores path and hops for
undirected graph in large dataset subgraph of size k longer than shorter path
Table 1 Comparison
IV. CONCLUSION Network”2020 11th International Conference on

Information and Knowledge Technology(IKT)
The improvements show an effect on search cost and [4]. E.A. Akkoyunlu,”The enumeration of maximal
the efficiency of the search. To extract frequent patterns, FP- cliques of large graphs, ”SIAM Journal on Computing
tree algorithms combined with LRU structure, UPTree is ,vol. 2,no. 1,pp.1-6,1973
exploited. It modifies the probability of selecting actions [5]. M.Regneri,”Finding all cliques of an undirected
based on degree of nodes to improve the selection process. graph,”2007
Top-k betweenness centrality nodes estimated to flow the [6]. Peter Mika. “Ontologies are us: A unified model of
information all over the network. social networks and semantics”, In Proc.14th
International Semantic Web Conference, 2005
V. FUTURE SCOPE [7]. Matsuo Y, Hasida K, Tomobe H, Ishizuka M.
“Mining social network of conference participants
In this paper, we discussed the algorithms which are from the web”, IEEE/WIC International Conference,
mainly based on undirected graphs for finding required 2003
information which can be found by one side only. So we can [8]. Kazuki Nakajima, Kenta Iwasaki, Toshiki
upgrade the algorithm to use it in directed graphs and can Matsumura, Kazuyuki Shudo, “Estimating Top-k
suggest searching from both ends. Betweenness Centrality Nodes in Online Social
Networks”, International Conference on. IEEE,2018.
REFERENCES [9]. L. C. Freeman, “A set of measures of centrality based
on betweenness,” Sociometry, vol. 40, no. 1, pp. 35–
[1]. GU Yiran, ZHAO Wenwen,"Improved Search 41, 1977
Algorithm Based on Probability of Node's degree on [10]. T. Matsumura, K. Iwasaki, and K. Shudo, “Average
Complex Networks" in 2012 Third International path length estimation of social networks by random
Conference on Digital Manufacturing & Automation walk,” in Big Data and Smart Computing (BigComp),
[2]. Suwook Ha, Yong Mi Lee, Kwang Woo Nam, and 2018 IEEE International Conference on. IEEE, 2018,
Keun Ho Ryu,”An Algorithm for Finding Frequent pp. 611–614
Patterns in Social Media Stream”, in International [11]. B. Goethals, M. J. Zaki, Workshop Report on
Conference on ICT Convergence (ICTC) on 14-16 Workshop on Frequent Itemset Mining
October 2013 Implementations (FIMI), ACM SIGKDD
[3]. Mohamad Mehdi Daliri Khomami,Alireza Rezvanian, Explorations Newsletter - Special issue on learning
Ali Mohannad Saghiri, Mohammad Reza Meybodi, from imbalanced datasets, vol. 6(1), June 2004.
“Distributed Learning Automata-based Algorithm
for Finding K- Clique in Complex Social

Unpacking The Black Box An Exploration of The Algorithms Driving Social Media Networks-A Review

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Unpacking The Black Box An Exploration of The Algorithms Driving Social Media Networks-A Review

Uploaded by

Copyright:

Available Formats

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology

Unpacking the Black Box: An Exploration of the

Ansika Jaiswal Sapana Survase

Social media may help us connect with friends and

IJISRT23MAY444 www.ijisrt.com 819

IJISRT23MAY444 www.ijisrt.com 820

Probability of Node’s Finding Frequent Finding K- Clique

IV. CONCLUSION Network”2020 11th International Conference on

IJISRT23MAY444 www.ijisrt.com 821

You might also like