tensorkelechi/arxiv-cstext · Datasets at Hugging Face

text stringlengths 0 89.3k
foraging theory has not been addressed in the existing lit
erature Combining the IFT with an enhanced variant of
EHO can substantially contribute in addressing the problem
of large scale information access on social media
III Analogy between animal food foraging and
information foraging on social media
The information foraging process intends to find paths lead
ing to relevant information on the Web The theory behind
it is based on the analogy between animal food foraging be
havior and human information seeking behavior It assumes
that when searching for information online users follow in
dications and hints that guide them to relevant information
similar to how animals follow the scent of their preys to catch
them Figure 1 and Table 1 present a good illustration of the
analogy between information foraging and animals food for
aging
Figure 1 Analogy between Information Foraging and Food
Foraging
Elements Food Foraging Information Foraging
ActorsPredator User
Prey Relevant information
Trigger Hunger Information needs
Environment Nature wilderness Web structure social graph
Cues Scent of the prey Hyperlinks icons titles
Table 1 Food Foraging analogy with Information Foraging
The following subsections describe our proposed model for
adapting information foraging theory to social media plat
forms as well as the basic notions on which the analogy with
animal food foraging is groundedA Territory social graph
In the OFT it is assumed that each animal operates in a de
limited geographical territory within which it searches for
food In information foraging the territory corresponds to
the search space composed of information sources such as
documents images and Web pages When it comes to so
cial media the users shared content serve as information
sources A social graph 27 is a representation of the users
their shared posts and their social interactions and relation
ships
In this paper we model a social network as an oriented graph
GV E where
the set of vertices Vrepresents the social media users
the set of directed edges Erepresents relationships and
interactions in the network such as a post a repost a
friendship a mention a reply and a follow
A simplified social graph structure is shown in Figure 2 The
edges that reflect the relations post repost mention and re
ply contain the social posts and so represent the information
sources We denote the set of these contentsharing edges by
Ewith EE
Figure 2 Social graph structure
B Food diet users interests
Each animal in the food chain has its own preferences in
terms of food Wilde animal for instance choose their preys
based on their environment their size and their hunting skills
In information foraging the animal food diet is translated by
the users information needs that we call the users interests
The information foraging process takes two inputs a collec
tion of posts represented by a social graph in addition to the
users thematic interests These users interests can be ex
pressed explicitly by the user or inferred implicitly from the
users social media activity profile and interactions 14
The modeling of the users interests consists in extracting the
terms that are the most representative of the users informa
tion needs from the keywords given by the user andor the
information accessible on their profile biography previous
posts etc
The extraction process includes tokenization stop words re
moval stemming and Term Frequency TF calculation The
words with the highest TF values are then stored into the
users interests vector Ifollowing the bagofwords model
Figure 3 illustrates the process of modeling the users topi
cal interests by a vector of terms and using it in information
foraging31 Y Drias et al
Figure 3 Users interests extraction
C Scent information scent
The general goal of information access approaches is to of
fer mechanisms that can help finding relevant information
while minimizing the time spent doing the search Likewise
the goal of animals in the wild is to find a decent amount
of food whilst spending less energy To achieve that ani
mals generally rely on their senses to locate and hunt their
preys in an effective way The authors of the IFT notice that
users have a similar behavior when looking for information
on the Web The authors assume that when browsing the
Web users exploit available hints and cues to estimate the
information value contained in accessible pages and there
fore decide which pages to visit This can be achieved thanks
to the information scent concept 28 which can be seen in
real life as the users estimation of the value that a source
of information will deliver to them This value is primarily
computed based on the sources descriptioncontent In the
case of the Web for instance information sources are Web
pages which are described by a URL a title and in certain
cases an icon
The goal in our context is to find relevant posts to satisfy a
specific users information needs We presume that if a post
is related to the users interests it will be more appealing to
them and that the information scent value should increase as
we get closer to a relevant post and decrease otherwise We
define the information scent as the similarity evolution be

foraging theory has not been addressed in the existing lit

erature Combining the IFT with an enhanced variant of

EHO can substantially contribute in addressing the problem

of large scale information access on social media

III Analogy between animal food foraging and

information foraging on social media

The information foraging process intends to find paths lead

ing to relevant information on the Web The theory behind

it is based on the analogy between animal food foraging be

havior and human information seeking behavior It assumes

that when searching for information online users follow in

dications and hints that guide them to relevant information

similar to how animals follow the scent of their preys to catch

them Figure 1 and Table 1 present a good illustration of the

analogy between information foraging and animals food for

aging

Figure 1 Analogy between Information Foraging and Food

Foraging

Elements Food Foraging Information Foraging

ActorsPredator User

Prey Relevant information

Trigger Hunger Information needs

Environment Nature wilderness Web structure social graph

Cues Scent of the prey Hyperlinks icons titles

Table 1 Food Foraging analogy with Information Foraging

The following subsections describe our proposed model for

adapting information foraging theory to social media plat

forms as well as the basic notions on which the analogy with

animal food foraging is groundedA Territory social graph

In the OFT it is assumed that each animal operates in a de

limited geographical territory within which it searches for

food In information foraging the territory corresponds to

the search space composed of information sources such as

documents images and Web pages When it comes to so

cial media the users shared content serve as information

sources A social graph 27 is a representation of the users

their shared posts and their social interactions and relation

ships

In this paper we model a social network as an oriented graph

GV E where

the set of vertices Vrepresents the social media users

the set of directed edges Erepresents relationships and

interactions in the network such as a post a repost a

friendship a mention a reply and a follow

A simplified social graph structure is shown in Figure 2 The

edges that reflect the relations post repost mention and re

ply contain the social posts and so represent the information

sources We denote the set of these contentsharing edges by

Ewith EE

Figure 2 Social graph structure

B Food diet users interests

Each animal in the food chain has its own preferences in

terms of food Wilde animal for instance choose their preys

based on their environment their size and their hunting skills

In information foraging the animal food diet is translated by

the users information needs that we call the users interests

The information foraging process takes two inputs a collec

tion of posts represented by a social graph in addition to the

users thematic interests These users interests can be ex

pressed explicitly by the user or inferred implicitly from the

users social media activity profile and interactions 14

The modeling of the users interests consists in extracting the

terms that are the most representative of the users informa

tion needs from the keywords given by the user andor the

information accessible on their profile biography previous

posts etc

The extraction process includes tokenization stop words re

moval stemming and Term Frequency TF calculation The

words with the highest TF values are then stored into the

users interests vector Ifollowing the bagofwords model

Figure 3 illustrates the process of modeling the users topi

cal interests by a vector of terms and using it in information

foraging31 Y Drias et al

Figure 3 Users interests extraction

C Scent information scent

The general goal of information access approaches is to of

fer mechanisms that can help finding relevant information

while minimizing the time spent doing the search Likewise

the goal of animals in the wild is to find a decent amount

of food whilst spending less energy To achieve that ani

mals generally rely on their senses to locate and hunt their

preys in an effective way The authors of the IFT notice that

users have a similar behavior when looking for information

on the Web The authors assume that when browsing the

Web users exploit available hints and cues to estimate the

information value contained in accessible pages and there

fore decide which pages to visit This can be achieved thanks

to the information scent concept 28 which can be seen in

real life as the users estimation of the value that a source

of information will deliver to them This value is primarily

computed based on the sources descriptioncontent In the

case of the Web for instance information sources are Web

pages which are described by a URL a title and in certain

cases an icon

The goal in our context is to find relevant posts to satisfy a

specific users information needs We presume that if a post

is related to the users interests it will be more appealing to

them and that the information scent value should increase as

we get closer to a relevant post and decrease otherwise We

define the information scent as the similarity evolution be