Professional Documents
Culture Documents
ISSN No:-2456-2165
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.
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
III. COMPARISON