Huddersfield-paper

Data Visualisation in Art, Design and Science Collaborations
deadline: 16.02.11

University of Huddersfield, UK, March 10th-11th 2011

site de présentation de la conférence

second draft waiting for review ( 06.02.11)

Buzzaar, An Environment and Tool to Map Your Web Keywords: web, collaboration, data visualization, mapping, user, knowledge sharing, network

structure of the text
1. ABSTRACT

2. INTRODUCTION -- dEfinition of goals, main principles, general context

3. SECTION 1 CONCEPT -- what and why is developed through the project

a 2 populations of users guest and registered users (differences and commonalities)

b. main principals behind the interactive graphical visualization of web activity data :notion of webmap; functionnalities to manipulate the webmap (tagging, filtering customizing,merging projecting)

c usecases scenario

4. SECTION 2 EXPERIMENTAL PROTOTYPE --the buzzaar server (+ illustratif: screeshots site web, toolbar, customisation, tagging, filtrag

5. SECTION 3 RELATED WORKS

6. CONCLUSION

7- REFEREMCES -- bibliography webliography

AUTHORS
Nathalie Perrin, meetopia- netart, Lausanne, perrin@meetopia.net

Katarzyna Boron, meetopia- netart, Genève, cax@meetopia.net

Adla Isanovic, Academy of Fine Arts Sarajevo, adla_isanovic@yahoo.co.uk

Alexander Elmer, Lausanne, alexander66@sunrise.ch

Martin Rajman, EPFL Lausanne, Switzerland, Martin.Rajman@epfl.ch

Tran Huu Duc, EPFL Lausanne, Switzerland, huuduc.tran@epfl.ch

ABSTRACT
As Web users, we visit pages and our navigation history is often locally stored in our browser logs thus creating a data set that contains a lot of valuable information, not only for ourselves but also for peer users, we might be willing to share this information with.

How can we best exploit this information? How can we give it a meaning and transform it into knowledge that  can be shared with others?

Buzzaar is a tool and a web site that provide answers to such questions. It is free and available for all. Buzzaar users can interact with their browsing history, and transform it into meaningful, dynamic webmaps that intuitively and dynamically represent their interests and navigation habits.

A webmap is an efficient representation of the web pages visited by a user. It can be easily modified and tailored to specific user needs. In particular, users can tag, filter, and sort the visualized data, and customise its appearance.

Webmaps can also be merged or projected into each other to identify commonalities or differences, thus allowing their owners to better understand and exploit  the knowledge they contain.

The Buzzaar website also allows anyone to register to the Buzzaar network, within which peers can share their webmaps and collaboratively build a common webmap representative of their community as a whole. This common webmap provides a useful context to better interpret each peer's individual webmap.

Info: hhtp://www.buzzaar.net

INTRODUCTION
The development of digital technologies has  brought fundamental changes into our lives, culture and society. Social networking media (i.e. Twitter, facebook),  user generated sites (i.e blogs, Wikis, YouTube, Flickr, delicious…), and peer-to peer networks[1] have contributed to build a new participative culture where “users[2]”  have access to tools to create, share, remix contents with peers.

As web users,we spend significant time online, searching for contents and visiting numerous web pages. These actions leave traces that can be collected, typically within browsing history logs that contain visit frequencies , and if the specific buzzaar browser plugin ( hereafter called the buzzaar toolbaar) is used , visit co-frequencies. These data represent potential sources of valuable information for various, private and public purposes.

Commercial companies (e.g search sites such as google, bing, yahoo …, e-commerce sites such as amazon or specific purposes sites such as www.compare.com) have  well understood  how useful  the above mentioned data   can  be when properly analyzed and transformed into exploitable information  about their targeted audiences. However, in most cases, users cannot access the generated information that is kept private although the original data have been created by the i web users, themselves.

On the other hand, a lot of personal navigation data is locally stored in users' browser logs (navigation history), but there are currently no popular and easy to use tools to exploit it and they are quite difficult to share in an efficient way. (e.g sites such as Delicious do allow to share bookmarks but do not provide any intuitive graphical visualization nor they provide mechanisms for users data aggregation).

Besides nowadays societies produce  such an amount of data that it has become difficult to   to read, understand,  and exploit  information from it. The visualization of data allows to visualize this information and then give more sense to it.

It show relations that are otherwise difficult to represent. Tools for visualization create dynamic interactive displays where users can see, explore, sometimes even touch and manipulate contents. They are also tools for contextualization, conscientiousness, and actions and can potentially contribute to public awareness and lead to their empowerment.

Data created by users should return to users who should be able to interact with it and  - if they wish -  to share it with others, in order  to  build a culture of  the commons [3]; a participative and distributed knowledge, part of our digital culture .The goal of the buzzaar project described on this paper is to fill the identified gap. In particular, the project aims at building a free and publicly accessible server (the buzzaar server) providing:

-a firefox plugin ( the buzzaar toolbar) allowing the users to gather their site visit frequencies and co-frequencies and to store them in a private space on the buzzaar server;

-a java applet ( the webmap manager), running on the buzzaar server, allowing registered users (1) to visualize their navigation data in the form of interactive graphical webmaps that intuitively represent their interests and navigation habits.; and (2) allow them to share their webmaps and collaboratively build a common webmap representative of their community as a whole; This common webmap (as any other shared webmap) can then be used by all registered users to be merged or projected into their private webmaps in order  to identify commonalities or differences, thus allowing them to better understand and exploit  the knowledge they contain. In addtion, the common webmap is also accessible to any (not necessarily registered) user through the buzzaar server, and, as such, represents a valuable source of public information, when manipulated with the webmap manager.

Both the buzzaar toolbar and the webmap applet are under the GPL license and the content of the buzzaar website are under a creative commons license to contribute to the development of a common culture on the web.

The long term goal of the buzzaar projects is to contribute to the development of a widely exploitable set of web data management techniques and tools allowing reliable collaborative creation of publicly shared information and providing efficient ways to protect users' privacy by keeping their identity and contributions anonymous. The team of the buzzaar project is composed of researchers from the  EPFL Artifical Intelligence Laboratory   and the netartists from the meetopia community[4]. The rest of this paper is organized as follows: part 1 deals with concepts… ; part 2 deals with experimental buzzaar prototype; part 3 deals with related works and finally part 4 provides some general concluding remarks.

1 - CONCEPTS
Two populations of users: guests and registered users, differences and commonalities

Buzzaar is a tool that is meant to be free and accessible for all: everyone can use it. It provides users with two different spaces: The buzzaar common webmap which is the result of all merged data from all registered users and a private space for each member, where they can manage their data stored on the server.

Buzzaar distinguish two types of users: the guests and the members. Each of them have a different status.

The guests can play with the common webmap. They can filter, tag, project , merge, export /import  their given results.

Members have different privileges; they have their ownprivate space where they send their data  They can exploit, manage and store their .webmaps. They can use all thefunctions provided by the application. They can access and interact with the buzzaar common webmap, which provides a useful context to better interpret each peer's individual webmap.

In order to have their own map and interact with it, to send their data to the server, users need to (1) register to buzzaar, (2) download and install the buzzaar toolbar, which allow them to send their data to the server. (send = vote for an URL)

The buzzaar toolbar is a plugin for Firefox web-browser. This tool helps users to manage their data (gather and send) from their browser history to the buzzaar server. Directly from the plugin, members can define if they want to ban some url sharing (avoiding to send sensitive and personal information to peer review). At any time, they can decide either to start sharing (send) or stop sharing (suspend from sending) their data to the buzzaar network.

When peers vote for an URL they send it to the server; it means they also add it to the buzzaar common webmap therefore they share their data with peers and contribute to build a common knowledge.

main principles behind the interactive graphical layout visualization

Buzzaar webmaps are visual representations of the rough textual data stored on the server.and sent by its members.

There are several examples of webmaps that efficiently visualize relations between web pages visited by web users. (SEE related works.). They can translate huge amount of data into  visual representations that are simple and understandable.

Since today’s maps can be fully interactive, multimedia, explorable, collaborative and sharable. Users are in position to experiment with their emotional and intellectual understanding and knowledge building.

One well-known figure for visualizing dynamic data is the web mapping.

For these reasons, we have chosen dynamic and interactive web graphs that users can edit, manipulate, filter and share, transforming their data into understandable visual representations which can be easily modified and tailored to specific needs and interest. Operations such as filtering, searching, merging and projecting results can be easily monitored and provide users with a different display of information at each time. Therefore, webmaps have infinite display possibilities: They produce graphs with up to date data, they can filter information, aggregate contents, enable tools for layout personalization.

They show different informations such as pages popularity (number of page /URL visits), co-frequencies (relational frequencies among pages and/or users), categorization (tagging and filtering), using different visual artifacts: size, opacity, position, colors and alpha transparency.

Therefore, in acting on their webmap and with the different visual elements, users can easily see and understand differences among pages, nodes or similar data; they can associate meanings and better situate themselves in the informational context.

Webmap manipulation (tagging, filtering customizing,merging, projecting)

Aggregation

Webmaps contains different levels of information: domain URL -associated   url (files and sub-domains) - and nodes  -files of different type  - images, documentation, movies, web pages.

Webmaps have three different aggregated1 modes: domain mode tag mode and co-frequencies.

The domain mode is the layout that provide users with their history of navigation classified by domain URL.

The tag mode is the layout that provide tag classification (see below tag and filter) The aggregated domain has visual markers that signify their nature. User can visualize information at different levels of granularity2 (e.g. aggregated level or detailed level) by expanding or collapsing aggregated node under examination.

Aggregated contents such as co-frequencies[3] help contextualizing user's interests. It can be used as requests to a peer for a recommendation or advice (if they share same interest), Many companies ( such as Amazon) already use this type of collaborative filtering and recommendation. Results as recommendations and suggestions are mostly rendered in textual way.

Users can decide whether they want to see co-frequencies and how (all co-frequencies, co-frequencies from or to activated node). That action shows users what relation they make between data and how they behave online with informations.

Tag and filter

When surfing on the web, users may want to collect and classify information they find important. Each web browser provide them with the possibility to tag and bookmarks their favorite links. t Bookmarking links has become a very popular behaviour. Social bookmarking web services such as Delicious offer the possibility to its members to share their bookmarks with their peers.

Buzzaar provide users with a tag mode, allowing them  to classify their links and give  a personal meaning to their information: users can create as many  tags they want, differenciate them with colors. When shared with peers, a tagged webmap keep its tags, allowing users to share their tag as well as their data.

Users have also the possibility to filter their data in order to focus on specific topics or URL names. They can save the result of such requests as a separate map too (to save it as a particular moment or part of the research).

In order to help users in manipulating and interacting intuitively with their data, Buzzaar allow them to share their webmaps with their peers.(they can send their data to friends or relatives. ) They can upload  one peer's data, compare and add it to theirs in order  to manipulate and improve their results.

Two operations are designed for that. Merge and projection.

The merge is when users add someone's data to theirs.

In this process, resources of two maps are merged. The highest value counts for the whole. Itt allows usersto put their resources in common, and build a representation of common interests. mathematical formula:of the merge operation

Project

The Projection is when users compare someone's data with theirs.

This process represents a comparison and a differentiation of data. It allows users to see what kind of information from the projected map is related/similar to theirs and then eventually to add these information to their map by tagging and saving this result. The projection allow users to compare, access to others' resources fullfill their interests and improve their knowledge. Mathematical formula of the projection operation

Webmaps representation and customisation

Webmaps are user-centered. The Graph does not represent a network without a center. It represent a map of users' interests and knowledge and it needs to be personally addressed. Users are at the center surrounded by their own navigation history.

The interface layout it divided in two parts: the webmap and the list. The webmap is the graphical translation of visited URLs, while the list is its textual representation. Basic principles that define the relationship between graphical environment and list are simultaneity and reciprocity - any change on the list is immediately represented on the graph and vice versa.

Nodes and domain URL  are represented by circles. Each of them represent one level of information. Circles are commonly used in other similar efficient graphics and visualizations (see related works). This metaphor is familiar  and easily understandable.

These -nodes - are displayed in a matrix divided into four circled layers, from the center to the periphery. Each of them represent a different area of  information: (1) The- central circle  (first circle)contains a visual marker for the user (an icon that can be customized); (2) The second circle is the area  for the expande  aggregated information (domain URL or tag); (3) The third circle contains the co-frequencies (I.e links related  to)  to the expanded information; (4) The last circle contains external data :co-frequent nodes of a projected map, tag names /images/

Structure and representation of the system ( architecture trust and privacy)

To come !

usecases scenario

2 -- EXPERIMENTAL PROTOTYPE
Working with their webmaps, users create new graphics, new environments containing useful information that can be shared with others. . //image//

When users want to tag their information, they have to enter the tag mode display (they leave the domain mode). All nodes that aren't tagged are gathered in the «un-tagged» category. When users create tags, they can define name and colors. When they tag their nodes, they aggregate them; their nodes move from the «un-tagged category» to the defined tag. //images//

As said before,. Merging two maps creates a new map that keeps the highest value of both I Example: friends want to create their common webmap, so they merge all their personal webmaps together, one by one. The result is their common map. Add image

Projection //iimage// Example: student “A” wants to compare her sources with student “B”. Student “A” will project her map into the map of student “B” and will analyze the result showing related nodes

In this comparison Co-frequent nodes from the projected map are displayed within third circle and have different color (in relation to the rest of the map)./image/.

RELATED WORKS
management). Some of them have also used graphs and visualization to map users navigation paths, bookmarks etc.

Apart from Buzzaar, there are other projects that already addressed the management, display and share of users’ bookmarks and history files (in terms of knowledge management). Some of them have also used graphs and visualization to map users navigation paths, bookmarks etc.

One of the projects that we found related to Buzzaar was Internet Cartographer (software) from Inventix Software (ref n1). It is a software made for older Windows operating system, and seems not to be ported anymore). This software installed on a client machine was meant to manage user's navigation history. It could manage bookmarks and had tools like research by keywords and co-frequencies visualization.However, it is a client software and did not address the issue of sharing information between peers.

Since 2009, Firefox users can install Voyage (ref n2), a Firefox plugin developed by Hsiao-Ting Yu. Voyage is designed to map and visualize navigation history, to provide users with an interactive tool that enables them to search into their history files. Hsiao-Ting Yu developed this tool because he was, as he explain, fed up with the long textual list of his firefox history log and since he wanted something more intuitive. He missed information on co-frequencies which could not be shown in the textual history log.

Voyage's users "travel" visually inside their history navigation map throughout a timetable. They can select a specific date, can see frequencies, co-frequencies on nodes information and Media Wall that displays multimedia files visited by. They can also integrate their twitters' tweets.

Delicious is another social bookmarking website for storing, sharing and discovering web bookmarks. It offer the possibility to its members to share their bookmarks with their peers. using non-hierarchical classification system of taging bookmarks with freely chosen index terms (generating a kind of folksonomy).

CONCLUSION
As web users we spend significant time on line searching for informations, webpages … these actions creates data that could be interesting and usefull for everybody if only we could easily access to it and give it a sense: I.e transform it in a meaningfull knowledge that we could share with peers.

These data are stored partially in our browser logs as our “navigation history” but it is hard for average users to understand and take something out of these rough textual data.

Data created by users should return to users in a form of a common knowledge that everybody can use and refer to. Each of us should be free to manipulate it, share it at their whish.-

We have created buzzaar exactly to respond to these challenges. We have designed a tool that not only transform rough data into meaningfull visual graphs but that allow users to fully interact with it.

Users can manipulate their data, filter it tag it and share it with peers.

More; Buzzaar provide the wide communtiy of web users with a public webmap made of all the links from all buzzaar users as a common, free and shared knowledge at free disposal.

One challenge that still need to be faced is the populartiy of the tool: the more people will use it the more interesting results will be.

We feel concerned with users privacy and anonymity. We don't want our tool to be used by those who would like to profile users for private needs. We have worked with the engineers of the LIA of EPFL to provide efficient answers to this issue.With their competences, we have build a tool good enough and well responding to the criteria of trust and privacy.