The double handshake mechanism:

1^st handshake: Requester - Server

The requester submits his match request to the server, which returns a list of possible matches, showing: (a) the degree of separation between requester and recipient and (b) a short profile summary describing the recipient.

It is up to the requester to select the best profiles before the match request is sent to the recipients. This allows for flexibility and accuracy, since the Requester has full control over the distribution of his request, and also handles the complex matching mechanism.

2^nd handshake: Recipient - Requester

The request is broadcast to matching Recipients. The request offers the recipient both the content of the request and the degree of separation between him/her and the Requester. Taking into account this information, the Recipients may choose to answer the request (or not). Again, this allows for flexibility and accuracy since recipients have the power to decide whether they know enough about the content of the request as the well as the requestor to reply (or not).

If the Recipient accepts to answer the request, the second handshake is successful and the match is complete. The Requester and Recipient can start exchanging messages.

F. SmallWorldz Architecture

The system is built from/using six distinct software components:

 - The Jabber Server [13]: Handles instant messaging between users.

 - The SmallWorldz Modules: Connect to the Jabber server to provide the SmallWorldz service.

 - The Database: To store all user data.

 - The Client: The main user interface with SmallWorldz. It allows users to log in and out of SmallWorldz, to send or receive requests and exchanges messages with others.

 - The Website [14]: It allows the users to register, manage their account and download the SmallWorldz client.

 - The Installer is user-friendly and fully automated software used to install the client on a local computer. The installer, provided by ZeroG [15] needs to be downloaded from the website.

4     Client Software and Self-Assessment Capabilities         

A.    The SmallWorldz Client Software

     The SmallWorldz client prototype is the interface between the user and the SmallWorldz server. To provide you with a clear idea of the way the system works we will describe, in a stepwise manner, the matching process. (Steps that are an optional part of the matching procedure are not assigned a number).

1.  The user starts the SmallWorldz client and inputs a unique identifier, a personal password and the type (termed "topic") of community of users he/she wishes to join. See Figure 6 (a).

2.  When a user logs on to the SmallWorldz server, all of the sub-communities (termed "sub-topics") of the chosen community are displayed in the SmallWorldz client. See Figure 6 (b).

- New users (as well as, on occasion, existing users) are explicitly asked for their address book. If the user agrees to provide some/all of his/her e-mail addresses to the SmallWorlz server, then he/she is provided with a tool to select those addresses that he/she wishes to release to the server (only for SmallWorlz promotion). See Figure 6 (c).

- The "roster" window in Figure 6 (d) provides the user with a way of creating and editing a personalized list - actually tree - of acquaintances that he/she has made on SmallWorldz.

3. The user may also create or edit his/her own profile within the SmallWorldz community. To do this, the user selects a number of subtopics and chooses the criteria that best suit him. By doing so, the user accepts to act as, at least an occasional, Recipient for the requests of others in those subtopics of the greater community. See Figure 7.

4.  This profile is public and is stored on the SmallWorldz server. Other users - referred to as Requesters may send their match requests to the SmallWorldz server. See Figure 6 (b).

5. In reply to the Requester, the SmallWorldz server returns a list of possible matches -seeFigure 8. The list displays:

      i. The current status of the Recipient (online or offline);

      ii.The degree of separation between the Requester and the matched Recipients;

      iii. A profile summary describing each of the Recipients.

6.  The SmallWorldz client displays the list of Recipients to let the Requester scroll through it and select the type of people he/she wishes to attempt to contact. The degree of separation gives the requester a sense of proximity, which is hoped, will encourage the exchange. See Figure 8.

7.  Once the Requester has chosen the Recipients, he/she sends them his request directly. See Figure 8. The Recipients receive the exchange request, which contains the degree of separation and a natural language message from the Requester. It is up to the Recipient to decide whether or not to accept the exchange, if he/she does, the exchange can take place through instant messages (or later, via other means, external to SmallWorldz, such as e-mail or the public telephone network). See Figure 9.

             Figure 6 (a)                    Figure 6 (b)

           Figure 6 (c)                             Figure 6 (d)                                                                                                                            

Figure 6 (a, b, c, d). Views of the SmallWorldz client

Figure7. Subscribing to Categories within a Community

Figure 8. List of Recipients returned by the Server

Figure 9. Recipients Receive the Request and Start Exchanging with the Requester

B. Automatic Self-evaluation of System Usage

     The next step is measuring system usage, in terms of: basic demand for services; utility of social proximity; mode of use; and effective promotion. In the sequel, we present a sequence of important operational questions, followed by a number of especially designed (and implemented) statistical metrics. These statistics are useful in providing empirically based answers to the questions.

     The system does all of the data collection, compilation and charting automatically.

·   Does the system attract demand, over time?

     One statistic (figure 10) counts the number of connections. A rising number of connections demonstrates the increasing popularity of the system, whereas a decreasing number of connections indicates diminishing user interest in the system. If the system helps people get answers to their requests, connections should rise together with the number of new subscribers.

Figure 10. Number of Connections per Month

·   Does human proximity encourage online exchanges?

     The use of a matrix, as in figure 11, would show the correlation between the number of exchanges between two users and their proximity. In other words, does proximity encourage exchange? And does the average number of exchanges between people go up as a function of proximity?

Figure 11. Relationship between Proximity and Exchanges

·   How do the users actually use the system?

     A useful statistic (figure 12) would show if the system is used mainly for matching people or just for its instant messaging capabilities. In other words, are people sending requests and trying to find people; or are they just chatting with their friends?

Figure 12. Number of Search Requests per Month

·   Are people willing to promote the system by releasing some of the addresses they have?

     A useful statistic (figure 13) would show the number of times the users submitted their address books to the system. It would also show if the users are reluctant to submitting their contact lists. SmallWorldz does not assume that a great number of users will submit their address books to the server since only a few will suffice to grow the system.

Figure 13. New Address Submissions

5     Summary and Future Work

     Online exchanges between Internet users will definitely take on a number of new dimensions in future years. Today, there is a lack of tools that facilitate human networking over the Internet. The system we propose is a step towards realizing such tools. Its actual success depends on many pragmatic elements besides its functional characteristics and technical robustness. However, we are focused in this study

on explaining the conceptual foundations of our system as well as its technical design and operation. Field tests, already underway, in Montreal and Brussels, are using SmallWorldz self-assessment capabilities to provide empirically-based answers to questions about the system's feasibility as a real-world product (as opposed to a research prototype).

     SmallWorldz is built on the basis of two fundamental principles: (a) people are a most invaluable source of information, and other goods, potentially available via the Internet. Hence, locating the right person(s) is usually a much more profitable and efficient enterprise than searching for mute documents; (b) people need a trust-bridge, of some form, which would encourage them to connect with each other. In SmallWorldz, we use degree of separation, which is a simplistic, but reasonable reflection of social proximity, as a means of encouraging communications and hence exchange between apparently alien individuals.

     It is our belief that people are willing to seek other people; they just need a flexible, effective, and non-intrusive method that would allow them to engage in this endeavour. SmallWorldz attempts to realize this by: (a) making sure that people have control over how they are contacted; (b) ensuring that people provide profiles, which enhance the accuracy, and hence efficiency, of the matching process; and (c) retaining all private information, on users, on the central secure server.

     If SmallWorldz is to satisfy its mission of connecting people, many people, then it must have a means of growth. SmallWorldz grows by explicitly asking for the address books of SmallWorldz current users as well as new subscribers. Thereafter, these addresses are used to invite new people to join the community, and hence enrich it with their resources.

     We envision that the system can be used to create small communities in which users are willing to meet others for online exchanges. Examples of such communities include: (a) researchers looking for potential collaborators in a given field of study; (b) university students looking for other students interested in joining them in launching a new student society; and (c) single men and women seeking partners for life or an adventure journey oversees!

REFERENCES

[1] F. Hattori, T. Ohguro, M Yokoo, Matsubara, and S. Yoshida. Socialware: Multiagent systems for supporting network communities. In CACM, 42(3):55-61, March 1999.

[2] Walter van de Velde. Co-Habited Mixed Realities. In Proceedings of the IJCAI'97 Workshop on Social Interaction and Communityware, Nagoya, Japan, August 1997.

[3] H. Nakanishi, C. Yoshida, T. Nishimura, and T. Ishida. FreeWalk: Supporting Casual Meetings in a Network. In  Proceedings of the Conference on Computer Supported Cooperative Work, CSCW '96. ACM Press: New York.

[4] Bin Yu and Munindar P. Singh. A multi-agent referral system for expertise location. In Working Notes of the AAAI Workshop on Intelligent Information Systems, pages 66-69, 1999.

[5] S. Milgram. The small world problem. In Psychology Today 2, pp. 60-67, 1967.

[6] "The Small World project": Department of sociology, Columbia University, New York:  http://smallworld.columbia.edu

[7] Duncan J. Watts.  Six Degrees: The Science of a Connected Age.  Norton, 2003.

[8] J. Kleinfeld. Could it be a big world after all? The six degrees of separation myth. Society, April 2002.

[9] J. Kleinfeld. Six Degrees of Separation: An Urban Myth? Psychology Today, April, 2001

[10] AskMeCorporation: Network your employees' knowledge: http://www.askme.com

[11]AllExperts: Free Online Question and Answer service: http://www.AllExperts.com

[12] Experts Exchange. IT solutions and experts: http://www.experts-exchange.com

[13] Jabber Software foundation: http://www.jabber.org

[14] SmallWorldz © website: http://www.smallworldz.net

[15] ZeroG: Multiplatform software installation and updating: http://www.zerog.com

[16] T. Ishida. Towards CommunityWare. In Proceedings of the second international conference and exhibition on the practical application of intelligent agents and mutli-agent technology. Pp. 7-21. 1997.

[17] Yoshida, S., Kamei, K., Yokoo, M., Ohguro, T., Funakoshi, K., Hattori, F. Visualizing potential communities: a multiagent approach 1998. In Proceedings. International Conference on Multi Agent Systems, 3-7July1998. Page(s): 477 -478.

[18] Fukuhara, T., Chikama, M., Nishida, T. Designing large-scale communityware: a case study in the Public Opinion Channel. In Proceedings of the IEEE Workshop on Knowledge Media Networking, 10-12 July 2002. Page(s): 120 -125.

Page 1, Page 2