Visualizing and Managing Folksonomies

Antonina Dattolo { Emanuela Pitassi

University of Udine, Via delle Scienze 206, I-33100 Udine, Italy

June 25, 2011

Outline

Background

Open Issues

Our Proposal

Folkview: the formal model

Folkview views and authoring

Related works

Conclusion and future work

Background

I The advent of Web 2.0 shifts the task of classifying resources

from a reduce set of experts, to the wide set of Web users.

I Free classi�cations are made by user thanks to tags, which

generate a folksonomy (Vander Wal, 2004; Mathes, 2004)

I Several social tagging systems such as Bibsonomy (Hotho et

al, 2006) or delicious allow users to classify resources and

generate folksonomies, but they are di�cult to manage,

modify and visualize in dynamic and personalized ways.

Background: some examples

I The creation of personalized views, which may display alimited, well de�ned and personalized sub-portion of anentire hyperspace has already been considered in di�erentsettings.

I Adptive bookmarking systems, e.g.:I PowerBookmarks (Li et al, 1999)I Siteseer (Rucker et al, 1997)I Web Tagger (Keller et al, 1997)

I Start pages on Web browser, e.g.:I Netvibes (http://www.netvibes.com.it)I My Yahoo (http://my.yahoo.com)I iGoogle (http://www.google.it/ig)

Open Issues

I Social Tagging Systems su�er from di�erent issues:I the lack and the exigence of general methotodologies for

extracting semantic information (Dattolo et al., 2010)I the lack and the exigence of personalized and dynamic

workspaces in wich usersI Can visualize personalized views of the folksonomyI Can apply personal changes

I A crucial task for developers of current Web application is

how to model and create speci�c tools for providing

personalized views to the users.

Open Issues

I A folksonomy is usually represented by a tripartite graph or

network Various researches have dealt with the issue related

to the complexity of the nature of the graph itself, projecting

a folksonomy on simpli�ed structures (Lambiotte, 2006;

Dattolo, 2011).

I Generally a folksonomy is represented by a tag-cloud, but this

kind of visualization is not su�cient as the sole means of

navigation (Sinclair, 2007).

I Possible multiple visualizations of a folksonomy allow users

to:

I have a more e�ective comprehension of the semantic relationsof a folksonomy

I have a more useful navigation through the involved elementsI manipulate the existing relations among tags and resources

according to the user needs.

Our Proposal

I The main aim of this work is to propose and describe a novel,

distributed, modular system called Folkview, whereby a

folksonomy is conceived dynamically through the use of

multiple agents.

I These agents will be capable ofI managing the structural and semantic properties;I cooperating for obtaining common objectives;I o�ering personalized and dynamic views.

I Steps:I De�nition of the Formal Model which exploits multi-agents

systemI Personalized views Folkview Prototype

The formal model

I Traditionally, given the sets of users U, tags T and and

resources R, a folksonomy is de�ned as the set of tag

assignments

(ui ; rj ; tk) 2 U � T � R

where i = 1; : : : ; jUj; j = 1; : : : ; jT j; k = 1; : : : ; jRj, each of

them indicating that user ui has tagged the resource rj withtk .

I User pro�les, functions, metrics or semantic relations among

users, tags, resources and tas are not intrinsic properties of

the folksonomy.

The structural components of the folksonomy

I In order to de�ne a F , we identify three classes of sets:I Tui ;rj 2 T is the set of tags used by ui on rj ;I Rui ;tk 2 R is the set of resources tagged by ui with tk ;I Utk ;rj is the set of users that tagged rj with tk .

I Each set represents a structural component of the folksonomy,

and we call it structural ; the tags are grouped associating to

them a semantic label for identifying their meaning in that

dimension.

Our representation of a folksonomy

Figure: 6 structural dimensions (left) and the corresponding folksonomy(right)

The �gure above represents three linear paths that contain the

resources tagged by user u1, using respectively t1, t2 and t3.The labels associated with them are respectively u1; t1, u1; t2 and

u1; t3, and represent a sub-portion of her personomy.

De�nition of Structural Dimension and Static Folksonomy

De�nition

A structural dimension is a labeled path

Dui ;rj = (V ;E ; �)

where

I V = Tui ;rj is the set of vertices,

I E is the set of edges,

I �(e) = (ui ; rj) 8e 2 E is an edge labeling, and

degree(tk) = 0; 1; 2 8tk 2 Tui ;rj .

In particular, degree(tk) = 0; 1; 2 only if jTui ;rj j = 1.

De�nition

A static folksonomy F is a labeled multigraph given by the union

of three families of structural dimensions.

F =[

i ;k

Dui ;rj [[

i ;j

Dui ;tk [[

j ;k

Dtk ;rj

where ui 2 U, rj 2 R and tk 2 T .

A dimension as an agent

We can introduce the de�nition of the dimension Dui ;rj based on

the structural dimension Dui ;rj .

De�nition

A dimension D = (Ts;En;Re;Ac) is an agent where

I Ts = Dui ;rj ;

I En = fui ; rj ; t1; : : : ; tng;

I Re = f;g, initially;

I Ac = fadd-tag ; delete-tag ;modify -tag ; : : : g

Analogously, we can de�ne new classes of agent dimensions, not

only for structural dimensions. New dimensions can be created

directly from the user, or computed by the system applying speci�c

metrics, or generated applying ontological models:

I each dimension can contain other dimensions; dimension

associates a semantics to the set of grouped entities.

Folksonomy de�nition as a multi-agent system

De�nition

A folksonomy F is a multi-agent system formally described as a

labeled multigraph of agent entities, organized in semantic

contexts, called dimensions.

F =

n[

i=1

Di

All in a folksonomy is a computational agent, equipped with a set

of local variables, that de�ne its internal state, and a modular and

extensible set of procedural skills.

Folkview views and authoring

In the labeled multigraph contained in the de�nition of F we can

recognize the zz-structures (Nelson, 2004): they are

non-hierarchical, minimalist, scalable structure for storing,

linking and manipulating di�erent kind of data.

From these structures, we inherit many strengths, such as their

intrinsic capability to preserve contextual interconnections

among di�erent information, thanks to their particular properties.

Folkview views and authoringIn the �gure the presence of a black triangle symbol, in two

positions, correspond to selected/not selected : these triangles areassociated to scripts related to the session agent of the current

visualization, and represent the mean to interact with the

cell-agent.

When selected, the session agent asks to the chosen resource (r9,in our example) the set of actions Ac that can be activated on it.

Then r9 sends a multicast message to all the dimensions in

which it is included, and a run-time created contextual menu,

organized in three meta-categories (views, metrics and semantics)

is shown.I The �rst category menu is concerning the di�erent kinds of

possible viewsI The other two categories of functions o�ered by the menu, are

related to:I the computation of an extensible set of metrics,I the application of opportune semantic relations and ontologies

in order to generate, for example, speci�c recommendations oncontent, tag and user.

Related works

I Few works have addressed the problem of interactivevisualizations of folksonomies:

I customized cluster maps for visualizing both the overviewand the detail of semantic relationships intrinsic in thefolksonomy (Montero et al., 2007);

I using information visualization techniques (IF) to discoverimplicit relationships between users, tags and bookmarks,end-users have di�erent ways to discover content andinformation otherwhise di�cult to understand (Klerx andDuval, 2009)

I the TagGraph project (http://taggraph.com/) is a folksonomynavigator which visualizes the relationships between Flickr tags.

I Nevertheless these works do not provide neither

personalized views nor e�ective dynamic changes

according to the user needs or preferences.

Conclusion and future work

I We have proposed an innovative way to conceive a

folksonomy in terms of a multi-agent system, �rst de�ning a

formal model and then showing Folkview.

I We have built a partial, but modular and extensible,

prototype, based on a public dataset taken from delicious,

and that implements the structural aspects of the considered

folksonomy.

I As future work we want to extend the prototype:I to all the main functionality we discussed, focusing our

attention on a semantic personalization;I to extract data from a large number of social tagging systems.