Datasets:
tensorkelechi
/
arxiv-cstext

Dataset card Data Studio Files
xet
Community
1
text
stringlengths
0
89.3k
cial graph
Output SP a surfing path leading to a relevant post
1SP
2Locate the edge eicorresponding to the elephants initial
position xcij
3SPSP ei
4Ni
5for all adjacent edge ejtoeido
6 Calculate InfoScent ejusing Formula 1
7 ifInfoScent ej0thenNiNi ej
8 end if
9end for
10ifNithen return SP
11else
12 Select a new contentsharing edge to visit from Ni
following Formula 2
13 Go to 3
14end if
C Solutions evaluation
At the end of each iteration of the algorithm the solutions
the surfing paths fetched by the elephants are evaluated ac
cording to a fitness function To do so we compute the sim
ilarity between the users interests and each solution using
Formula 3
fxcij Simek I 3
Where
ekrepresents the last social post on the surfing path con
structed by elephant jin clan ci
Iis the users interests vector
Simek Irepresents the cosine similarity between ek
andI
D Updating Operator
The elephants positions are updated using Formula 4 at the
end of each iteration of the algorithm once the new solutions
have been evaluated
xnewc ijxcijαxbestc ixcijr 4
Where
xnewc ij is the new position of the elephant
xcij is the current position of the elephant
xbestc i is the matriarchs position33 Y Drias et al
α01 is an empirical parameter that defines the
influence of the matriarch over the new position of the
elephant j
r01 is a random number which aims at improv
ing the diversity
The position of each clans matriarch is also updated
throughout generations by utilizing Formula 5 to calculate
the average fitness of each clan After that Formula 6 is
used to compute the new matriarchs position using the posi
tion of the elephant with the closest fitness value to the clans
average fitness
favgc i1
ncinciX
j1fxcij 5
xnewbestc ixavgc iβ 6
Where
favgc i represents the average fitness value of the clan
ci
xnewbestc i represents the new position of the matriarch
of the clan ci
xavgc i is the position of the elephant with the closest
fitness value to favgc i
β01 is an empirical parameter which determines
the influence of the average position on the matriarchs
new position
nci represents the number of elephants in the clan ci
xcij is the position of elephant jin clan ci
E Separating Operator
The elephant with the lowest fitness value will leave the clan
at the end of each generation Formula 7 is used to create a
new elephant to replace the one that left
xworst xmin xmaxxmin 1r 7
Where
xworst stands for the position of the elephant with worst
fitness value
xminandxmax are the upper and lower bounds of the
positions interval
ris a stochastic and uniform distribution parameter
V Defining territories with clustering
Although elephants are not territorial animals they utilize
specific home areas during particular times of the year Their
home ranges vary from from 15 to 3700 square kilometers
24 to 5958 square miles depending on the population and
the habitat This delimited area helps elephants to better mas
ter their environment and remember the location of food and
water sources 31
Implementing this concept and incorporating it into EHOwould be of a great benefit to solve large scale problems In
fact dividing the search space into subareas based on some
problemrelated features can help to limit the search to one
of these subareas and thus improve the effectiveness and
efficiency of EHO
When constructing a surfing path during the information for
aging process the choice of the starting point can have a ma
jor impact on the outcome and therefore determine whether
or not the path will lead to relevant information In fact lo
cating the right post from which to initiate the navigation
with regards to the users interests is of a high importance
This becomes even more obvious and crucial when dealing
with large scale data as skipping posts that are not related to
the users interests could significantly improve the efficiency