Reflection on Shared Experiences - Evaluation of Similar Applications

Examensarbete

LITH-ITN-MT-EX--06/010--SE

Reflection on Shared

Experiences - Evaluation of

Similar Applications

Norzima Elbegbayan

2006-01-27

LITH-ITN-MT-EX--06/010--SE

Reflection on Shared

Experiences - Evaluation of

Similar Applications

Examensarbete utfört i medieteknik

vid Linköpings Tekniska Högskola, Campus

Norrköping

Norzima Elbegbayan

Handledare Bengt Lennartsson

Examinator Bengt Lennartsson

Rapporttyp Report category Examensarbete B-uppsats C-uppsats D-uppsats _ ________________ Språk Language Svenska/Swedish Engelska/English _ ________________ Titel Title Författare Author Sammanfattning Abstract ISBN _____________________________________________________ ISRN _________________________________________________________________

Serietitel och serienummer ISSN

Title of series, numbering ___________________________________

Datum

Date

URL för elektronisk version

Avdelning, Institution

Division, Department

Institutionen för teknik och naturvetenskap Department of Science and Technology

2006-01-27

x

x

LITH-ITN-MT-EX--06/010--SE

Reflection on Shared Experiences - Evaluation of Similar Applications

Norzima Elbegbayan

This thesis project aims to make a beforehand and empirical research for an online interactive tool that enables people communicate and share their thoughts and reactions on a shared event despite community size and geographic distribution. Existting online mass interaction systems and tools are evaluated, and potential users and organizers are interviewed.

Upphovsrätt

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –

under en längre tid från publiceringsdatum under förutsättning att inga

extra-ordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,

skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för

ickekommersiell forskning och för undervisning. Överföring av upphovsrätten

vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av

dokumentet kräver upphovsmannens medgivande. För att garantera äktheten,

säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ

art.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i

den omfattning som god sed kräver vid användning av dokumentet på ovan

beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan

form eller i sådant sammanhang som är kränkande för upphovsmannens litterära

eller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press se

förlagets hemsida

http://www.ep.liu.se/

Copyright

The publishers will keep this document online on the Internet - or its possible

replacement - for a considerable time from the date of publication barring

exceptional circumstances.

The online availability of the document implies a permanent permission for

anyone to read, to download, to print out single copies for your own use and to

use it unchanged for any non-commercial research and educational purpose.

Subsequent transfers of copyright cannot revoke this permission. All other uses

of the document are conditional on the consent of the copyright owner. The

publisher has taken technical and administrative measures to assure authenticity,

security and accessibility.

According to intellectual property law the author has the right to be

mentioned when his/her work is accessed as described above and to be protected

against infringement.

For additional information about the Linköping University Electronic Press

and its procedures for publication and for assurance of document integrity,

please refer to its WWW home page:

http://www.ep.liu.se/

Abstract

Most of the people attending a conference or workshop experience problems in meeting the right people to discuss their ideas due to reasons such as sociality, prejudice and inadequate experience etc. This thesis project aims to make a beforehand and empirical research for an online interactive tool that enables people communicate and share their thoughts and reactions on a shared event despite community size and geographic distribution. Exist-ing online mass interaction systems and tools are evaluated, and potential users and organizers are interviewed. Following the functional requirement architecture, a design prototype is outlined briefly. An original idea of a matchmaking tool/module of the system will be introduced with aspect of Natural Language Processing.

Acknowledgements

I would like to thank my supervisor and examiner Bengt Lennartsson for giving me a chance to work on this project, being patient with my failures and endless reasons, and encouraging me all the time.

I would also like to thank Jalal Maleki, the coordinator of International Master of Science Program, for his support in my studies at Link¨oping Uni-versity and for being a friend. Without his encouragements and advices, I could have given up anytime.

Lastly, I am very grateful for my husband Muno and my baby Michelle for giving up everything and accompanying me all the way to Sweden to help me finish my studies, my family for their emotional and financial support, and my friends, especially Irene Anggreeni, who backed me up every time I was on the edge.

Contents

1 Introduction 4 1.1 Background . . . 4 1.2 Purpose . . . 4 1.3 Problems/Tasks . . . 5 2 Related Research 7 3 System Characteristics and Study of Existing Applications 9 3.1 Methodology . . . 9

3.1.1 Identifying system usage . . . 9

3.1.2 Defining system characteristics . . . 9

3.1.3 Reviewing relevant applications . . . 10

3.1.4 Interviewing potential users . . . 10

3.1.5 System usage and behavior . . . 10

3.2 Scenario . . . 11

3.3 System Characteristics . . . 12

3.4 Practical Applications and Interviewees . . . 14

3.5 Interviewees . . . 16

3.6 Interactive Systems Evaluation Results . . . 17

3.6.1 Electronic mail (e-mail) and newsgroups . . . 17

3.6.2 Web forum . . . 18

3.6.3 Online chat rooms . . . 18

3.6.4 Instant messaging . . . 19

3.6.5 Multi user dungeons (MUD) . . . 19

3.6.6 Weblog . . . 20

3.6.7 Conclusion . . . 20

3.7 Online Dating Services . . . 22

4 Functional Requirements and A Design Prototype 25

4.1 Methodology . . . 25

4.2 Functional Requirements and Use Cases . . . 25

4.2.1 Scenario . . . 26

4.2.2 Use cases . . . 27

4.2.3 User registration and login . . . 29

4.2.4 Browsing and searching user profiles . . . 31

4.2.5 Browsing and searching discussion forum . . . 33

4.3 General or Non-functional Requirements . . . 35

4.4 Design Prototype . . . 36

5 Matchmaking Module and Natural Language Processing 39 5.1 Background . . . 39

5.2 State of the Art . . . 40

5.3 Information Retrieval (IR) . . . 41

5.4 Information Extraction (IE) . . . 41

5.5 Performance Level . . . 42

5.6 IE Tasks . . . 43

5.7 IE Systems . . . 44

Chapter 1

Introduction

1.1

Background

Imagine a conference, where hundreds of people, full of ideas and plans, are willing to share them with others. Most of the people experience problem in meeting the right people to discuss their ideas. The reasons are that some of them are people attending a conference for the first time and are ”lost” to some extent, some of them wait for others to come and engage in a conversation and some are looking for people with the same background and many more. They have to depend on their own personality, sociability and perhaps reputation. How can we build a system that helps people to find their ”match” during a conference without wasting their valuable time on trying to start a conversation with ”not so interesting” people?

1.2

Purpose

The main purpose of this thesis project is to make a beforehand research for an online interactive tool that enables people communicate and share their thoughts and reactions on a shared event (a lecture, a TV show etc.) despite community size and geographic distribution with possibility of idea matchmaking module1_.

As a starting point we will study existing online mass interaction2

sys-1_{The meaning of module could also be regarded as component.}

2_{Please notice the distinction between Interaction systems and Interactive systems. For} term ’mass interaction system’ please refer to [34].

tems and tools and evaluate them from usability point of view to extract system requirement specifications on ”shared reflection” for our anticipated system. During this stage we will hold interviews with some potential users and organizers with experience in conferencing and in the Internet. Further we would suggest a design prototype which combines existing system fea-tures and recommended improvements and also reflects the above mentioned specifications on ”shared reflection” and matchmaking.

Another purpose of the thesis is to outline a matchmaking tool/module of the system, which greatly involves information extraction and retrieval, branches of Natural Language Processing (NLP), as a basic survey. We should assume that the input data to our system are not only conference attendees’ personal and academic information but also their ideas and opin-ions. When one’s opinion is discussed, it is not straightforward to use query techniques like key phrase/string matching or indexing. Key phrase match-ing will result in ambiguous or vague matches if the input texts are on a same particular topic and are seen as bag-of-words. How NLP handles such idea extraction from text and further match them will be discussed in brief.

1.3

Problems/Tasks

We started our project by creating a list of problem statements and setting goals. What are the problems we want to solve? What are the goals we want to achieve? The list follows.

• Identify user expectations and system characteristics

• What kinds of systems and tools for mass interaction are available on

the Internet today? What features and functionalities do these tools provide and what advantages and disadvantages do they have?

• What would be the user requirements and expectations on a system for

sharing reflections on something experienced jointly (a conference)?

• What would be the added value for the organizer (school,

univer-sity, movie distributor, telecommunication operator, travel agent, etc.) when using such a support system?

• Develop a prototype according to the specified design requirements.

What are the motivations?

• What is the initiative of embedding idea matchmaking module? How

Chapter 2

Related Research

As my search shows not much was done specifically on how to match shared reflections on a jointly experienced event, which was our main project inter-est. However interactive systems that support virtual communities and their shared interest were ubiquitous.

ZENO [11], a mediating system with graphical interface and which was originally designed to support online mediation of discussions about political and planning issues [12], has a discussion forum component. The system acts as an intelligent assistant to human mediators, facilitators, arbitrators and other ”trusted third partie” by providing an issue-based discussion forum or conferencing system, building on decision theory and a formal model of argumentation. The approach is not limited to a particular school of argu-mentation, class of problems, or type of domain with its effort to provide variations and identify adequate parameters, both conceptually and techni-cally. Users place issues, positions and arguments into a ’picture’ using the graphical interface, which facilitates the browsing and retrieval of relevant past contributions to make the discussion richer, more precise and focused than the ”thread” mechanism typical of news groups. Using reason main-tenance and constraint satisfaction procedures, the preferences expressed in the arguments brought forward by the participants are summarized.

Delphi [31] is a method for structuring a group communication process so that the process is effective in allowing a group of individuals, as a whole, to deal with a complex problem. Usually Delphi undergoes four distinct phases. The first phase is characterized by exploration of the subject under discus-sion, wherein each individual contributes the additional information he feels is pertinent to the issue. The second phase involves the process of reaching

an understanding of how the group views the issue (i.e., where the members agree or disagree and what they mean by relative terms such as importance, desirability, or feasibility). If there is significant disagreement, then that disagreement is explored in the third phase to bring out the underlying rea-sons for the differences and possibly to evaluate them. The last phase, a final evaluation, occurs when all previously gathered information has been initially analyzed and the evaluations have been fed back for consideration. Delphi has been used in various applications, including evaluation of policy issues, identification of political events and analysis of the risk associated with these events, identification of software project risks, and development of an explanation facility in a knowledge based system [5].

Some of attempts to use computer mediated communication as conference or event supporting tools were made recently are McCarthy’s [20] use of IRC1

as digital backchannels in a shared physical space, an academic conference. As seen from his study on statistics of logs recorded during the conference and interviews with attendees, the digital backchannels were ”provocative and valuable, helping [the attendees] get acquainted with other attendee”, but having the backchannels simultaneously during the conference resulted in confusion and further inattention. Interactive Collaborative Environments Lab2_{at Swedish Institute of Computer Science}3_{setup mBlog (mobile weblog)}

to run live during the SITI Winter Conference 20034_{, enabling attendees to}

post an entry at any time from a mobile phone (MMS, SMS, e-mail, WAP forms, etc.) and to read at any time using a number of mobile outputs (WAP browser, speech synthesis). Similar research was done by Sumi et al. [27] which supported people to meet persons who have same interests and share information using mobile computers and web applications. Jacucci et al. [16] developed mGroup, a mobile group media application, that supports creating and sharing group experiences.

1_{Internet Relay Chat (http://www.mIRC.com) Oikarinen, J. and D. Reed, Internet} Relay Chat Protocol, RFC 1459, May 1993.

2_{http://ada.sics.se/newice/} 3_{http://www.sics.se}

Chapter 3

System Characteristics and

Study of Existing Applications

3.1

Methodology

The first purpose of the project was to identify our desired system’s usage and behavior and then assess existing practical applications similarity to them. After developing a list of tasks and requirements we moved on to evaluating existing applications.

3.1.1

Identifying system usage

We collected our ideas on why we need the system and started to identify user’s expectations from the system. Who are the system users and how will they use the system and what are their requirements? The actual need for the system arose with the planning of a particular conference; therefore the system is likely to be tested during this conference. Having this concrete idea and need, we began documenting possible uses and environments. We compiled a simple scenario expressing possible uses of the system and used this scenario to define user expectations and system behavior.

3.1.2

Defining system characteristics

We had identified our usage of the system and goals we want to achieve. Building on those assumptions we had outlined our system characteristics. Our system had to have several key characteristics and features in order to

achieve the goals initially set, and each characteristic could have a group of applications to assess correspondingly.

3.1.3

Reviewing relevant applications

We started by listing existing applications which we thought of having any similarity to the characteristics of our system. These applications were de-signed and used for diverse purposes and had different functionalities, for each of them was chosen against a single characteristic distinctly.

After brief review on the applications’ goals, tasks and usage, we assessed every application group’s similarities to our system characteristics and its performance result. Underlying technology and usability of the systems were also reviewed for later use in a prototype design and development process.

3.1.4

Interviewing potential users

During the evaluation process we interviewed several potential users with various backgrounds. The main intention of conducting these interviews was to explore the interviewees’ practical uses and experience with any kind of conference support tool. We also tried to discover possible design and usage expectations by real potential users.

3.1.5

System usage and behavior

The idea came about by a professor who is planning a conference which will take place in 2007 requiring the system to be implemented and tested during the conference. As an experienced conference attendee and planner he was exposed to some lack of concurrent interaction between participants. An interaction system for participants to share their reflections on lectures and meet other people regarding their thoughts and opinions was needed before, after and during the conference.

3.2

Scenario

A potential user is about to attend a big conference with more than hundred participants. There are series of lectures and seminars on various topics. The user will only attend lectures that are of interest to him. After a lecture or speech he would like to meet some people to discuss his reaction with, people who either might be working on the same project as him or have same reflection on the topic. A lecture usually involves large number of attendees and every single attendee could have different opinion on the topic. In series of lectures participants don’t have possibility to communicate with every other person and ask for their opinions. Therefore he decides to initiate communication and search for people to meet while he is at the conference.

There is an online system for sharing opinions and reactions of lectures or other events that will be (or was) held during the conference and the system is available on the conference web site. The web site contains (possibly) all information on the conference including list of lectures and events as well. People started sharing their thoughts and opinions well before the conference and in a wide scope.

The user goes to the web site and scans through list of events and lectures. (... ) Then he chooses the topic he would like to discuss his ideas on and by clicking in the link (lecture title) goes to interaction module. There he reads on lecture notes or abstract and some recent posts to get a general view of other people’s reaction. The posts have nested structure and a user can post either a reply message to one of previous message or a new comment. When a post is submitted he will have list of matching opinions sorted by their relevance to his opinion. The author of the first match in the list will be considered to have (almost) the same opinion as the poster. If he desires he can decide to initiate a communication with any of the recommended posters or just browse through to read all messages if the number of messages is reasonably small. Every post, including replies will have a link to a page, where their matching posts are displayed. In overview, a user can read posts by other visitors to the site, post a new post or comment/reply to a post, look for matching posts of any single post.

3.3

System Characteristics

As a result of the analysis on the scenario, the following list of characteristics was developed.

Available online

The system should be available on the Internet so users can access the system from anywhere. The advantage of a such system is that users could access it using not only computers but also various technologies including SMS (Short Messaging Service) and any other methods that browses the Internet such as WAP (Wireless Application Protocol). To make it possible for different users to browse the system from different browsers, the server and client technology might need more complex structures.

Interactive system

Users should be able to use the system interactively. Users will have access to interactive tools such as browsing through and posting a message etc. The system will respond to and take action depending on user’s behavior. How will interaction be handled between users and system?

Text message as communication mode

All communication will be performed by means of text messages. Size and content of the text is up to user’s preference. How is the information ex-changed between users and system? What could be possible alternatives and extensions?

Asynchronous system with centralized database

The interaction mode is supposed to be asynchronous. For the system to be able to match people using their data there must be centralized server/database and client architecture. Text messages should also be archived in the database and available for other people to read some other time. What is the advan-tage of being centralized? Is it possible to be decentralized? How is the information stored and retrieved in a database? What should we save in database?

Conference support tool

The original idea of the system is to support and enhance user interactivity during a shared event such as a conference. What conference support tools (not only for user interaction) are previously developed and tested and how is the user experiences?

Matchmaking and Dating

One of key features of our system is going to be ”matchmaking”. The system will perform a particular matchmaking action on text posted by users. The system should be able to help a user to find people with similar thought or opinion on a shared experience by matching messages and possibly arrange a date or meeting. What kinds of matchmaking do systems carry out? How do these systems do the matching? What techniques do they use? How efficient is the matching?

3.4

Practical Applications and Interviewees

”Share and discuss your ideas and opinions” systems

The list of applications to evaluate began with systems we use in everyday life; the ones we use to communicate with other people to share our opinions, ideas and experiences. The mostly and frequently used applications in our daily life are e-mail, instant messaging systems and possibly web forum. Web forum is discussed only when our thought is set to a particular topic. We listed every application and added similar products that we do not use ourselves but are exposed to generally.

We populated our list with the help of results from search engines to include as many systems as possible. While doing that we began to ap-ply our characteristics as keywords. The keywords were ”share/discuss your idea/thought/reflection, online interaction systems, text based communica-tion, computer mediated communicacommunica-tion, mass communicacommunica-tion, asynchronous communication, conference support tool, intelligent matchmaking/dating sys-tem” etc. with slight changes with their synonyms.

Available online

As a result of growing Internet technology, most of today’s interaction system are available online or at least use the technology (except telephone systems, of course). Therefore we did not have to specify the systems to be online during our search. However the systems we collected were both centralized (with server and client architecture) and decentralized (peer to peer). Interactive system

In order to express one’s reflection on a topic, users should be able to use the system interactively by posting data to the system and the system should respond to the them. These systems included chat rooms, newsgroups and online games excluding above mentioned applications.

Text message as communication mode

Possible modes of communication on the Internet are text, images, audio and video. For example, weblog and e-mail system use both text and images and contemporary chat rooms have audio and video conferencing possibility.

Asynchronous system with centralized database

Examples of asynchronous systems are e-mail and discussion forums. We, however, included some synchronous systems such as chat rooms and instant messaging, because these systems were big candidates of interaction systems and essentially they have databases, may be not for solely archiving messages but for other purposes.

Matchmaking and Dating

The term matchmaking usually refers to matching a pair of people based on their profiles for purpose of a romantic relationship and further initiating a meeting or date. We added the word ”intelligent” in search key phrase for matchmaking systems, we wanted to omit numerous online dating services and their scripts and include as many idea matching systems as possible in our search results. However the results were not encouraging for main tech-niques dealing with the intelligent matchmaking systems were about network ”crawling”.

Applications

As a result of our search we had the following list of applications to evaluate.

• E-mail

• Web (discussion) forum • Newsgroups

• Web blogging

• Multi user dungeons (MUD) • Instant messaging

• Chat rooms

• Classical) online dating services

The main questions we asked every application were: What is your goal as an interaction system? What tasks do you propose? What are your

advantages and disadvantages in communicating people? How can we use the advantages and avoid the others? Moreover, relevant questions were formulated for each characteristic.

3.5

Interviewees

The list of people to interview was compiled with the help of Professor Lennartsson, because he could use his social network to find a good po-tential user with adequate experience in conferencing while I did not have any formal connection with many people in Sweden. Following questions were asked.

• What is his experience of sharing his ideas on a conference (big and

small), how does he handle it?

• What is his experience using any support tools on a conference? What

are they?

• What does he think of such conference support tool? Does it seem

either a good idea or a duplicate? If it is a duplicate, what could make it original?

• What is his expectation from such a tool? What are his ground

re-quirements if he was the customer ordering the tool (to be developed)?

• Any idea on how else the tool can be used, other than in a conference?

(in his own case)

• If the tool is proven to be useful, what are the chances that it will be

successful on the market (such as institutions and companies)?

• Is there any (formal or informal) standard/requirement set by

3.6

Interactive Systems Evaluation Results

While evaluating applications and systems they were expanding our scope with more features and similar products. As a result we started to ask a new question: What is your new, advanced and promising feature (concerning our characteristics)?

3.6.1

Electronic mail (e-mail) and newsgroups

Electronic mail, abbreviated email or e-mail, is asynchronous method of com-posing, sending, and receiving messages over electronic communication sys-tems. While the basic technology is designed to pass simple messages between two people, even relatively basic e-mail systems today typically include in-teresting features for forwarding messages, filing messages, creating mailing groups, and attaching files with a message. These messages range from per-sonal messages to professional announcements, contributions to mailing lists, and even advertisements [33].

Newsgroups and electronic mailing lists are similar to e-mail systems ex-cept that they are intended for messages among large groups of people instead of one-to-one communication. In newsgroups messages are posted like they are on a bulletin board and membership is not required to view messages. In practice the main difference between newsgroups and mailing lists is that newsgroups only show messages to a user when they are explicitly requested, while mailing lists deliver messages as they become available. However, short-comings with electronic mailing lists are information overload and security problems such as spams and virus threats.

Both are ways of gathering group of people with similar interests and distributing news and information to such a particular group. Thus it can act as discussion and information sharing method among group of people who explicitly want to share their reflection on shared event. In addition to above mentioned drawbacks of e-mail systems, a large amount of work required to build and maintain both the list and necessary software, should be taken into consideration if such lists and groups are used. An alternative approach to mailing lists was suggested by Schwartz and Wood [24] using a heuristic algorithm to discover shared interests by analyzing e-mail communication history.

3.6.2

Web forum

A web forum is web application software for holding discussions allowing peo-ple to post messages and comment on other messages; and could be described as web version of newsgroups or mailing lists. Web forums are also commonly referred to as message boards, discussion boards, discussion groups, bulletin boards or simply forums.

One significant difference between web forums and mailing lists is that mailing lists automatically deliver new messages to the subscriber, while web forums require the member to visit the website, and check for new posts (same as newsgroups). The main difference between newsgroups and forums is that additional software is usually required to participate in newsgroups, a newsreader. Visiting and participating in forums normally requires no additional software beyond the web browser. However, a web forum system can have different newsgroups or individual mailing lists, and usually provides more than one forum, dedicated to a particular topic. Therefore forums (and newsgroups) are often arranged into hierarchies and implies newsgroups and mailing lists.

3.6.3

Online chat rooms

A chat room is a real time, online application where people can ”chat” by sending messages to people in the same ”room”. There are several differ-ent types of chat rooms: open chats, where people discuss anything; topic-focused chat rooms, where people discuss a particular topic, like a TV show or a sports team; and moderated chats that feature a speaker who leads a discussion. Some chat rooms are moderated either by limiting who is allowed to speak, or by having moderation volunteers patrol the room watching for disruptive or otherwise undesirable behavior.

Most chat systems are text based, where conversation is typed like text messages bouncing back and forward between a group of people in the same room and appearing immediately on their screens. Some chat rooms incorpo-rate audio and video based chat, which makes the conversation almost real. 2D or 3D graphical chat interfaces, such as The Palace [15] and Comic Chat [18], have been developed recently as alternatives to a text based chat. These graphical environments make use of avatars1_{, pictures, drawings or icons that}

users choose to represent themselves to convey social presence and identity [32].

Many web sites now have online chat rooms. Although the World Wide Webs initial protocols were not conducive to live interaction, the advent of Java has made Web-based chat rooms increasingly popular. The drawback of these chats is they only work with newer, Java-enabled browsers. There are sites which offer a wide variety of ”chat rooms”, or a page might offer just one room.

3.6.4

Instant messaging

Instant messaging (IM) is an interaction tool that allows users to communi-cate by sending text messages to each other. IM is almost real time com-munication with a fast network transmission. Users can also have access to advanced features like file transfer, audio and video chat and various online services and data such as media, news and weather report etc. However, each user has to use the same program to talk, and the programs are currently not compatible with each other.

Unlike chat rooms, users don’t join a room to start a conversation; instead they ”call” each other individually and exchange their text messages on one-to-one basis and several conversations could take place at the same time concurrently. Some IM systems allow multiparty communication, but it is still based on call individually model.

IM is generally Internet based and communications are client-server based. For example, the architecture of Jabber2_{, an XML based IM system, is}

ex-tremely similar to e-mail. However, whereas e-mail is a store-and-forward system, Jabber servers deliver messages in close to real time.

3.6.5

Multi user dungeons (MUD)

MUD was originally designed as online games in which players are involved in fantasy adventures that they create together as the game progresses [4]. Traditional MUDs are text-based systems where users connect to a central server. In a MUD environment, users can be logged on at the same time and giving commands to interact with one another. Once connected, they enter any number of different rooms, chat with other users in those rooms, and

create and modify artifacts in the rooms. Though primarily used socially, MUDs have been used to support collaborative work, though their text-only presentation has proved limiting. Advanced versions of the MUD, systems like Jupiter system [7] and the wOrlds system [28], enhance a traditional MUD with audio/video conferencing tools and shared whiteboards. Team-Rooms [23] does not directly support audio or video and emphasizes applets more suitable for real-time collaboration.

3.6.6

Weblog

A weblog, or simply blog, is a web site for writing messages on regular basis as a personal diary. Blog is set up using a specially-designed interactive tool and created and run by a single person, who is referred as an author, sometimes anonymously. The posts are usually text with links, but some blogs place emphasis on including other forms of media such as images, audio and video and every post is archived on the blog. Some other emerging forms of blogging are MP3-blog for in posting music from specific genres and mblog for ”mobile blog”. In a blog new posts are shown at the top, so visitors can read what’s new and they can comment on it or link to it or e-mail the author. There could be several categories in a blog and correspondingly several blog threads on a site.

As stated in [36], the difference of blogs from web forums or newsgroups is that only one person or group can create new subjects for discussion on their blog. A network of blogs can function like a forum in that every entity in the blog network can create subjects of their choosing for others to discuss. Moreover blog is easier to set up and maintain and express its owners attitude and point of view.

3.6.7

Conclusion

As seen from the Table 3.1, the most outstanding candidates similar to our system are web forum and weblog. However, weblog usually has one domi-nating author and reflects only his opinions. Furthermore the design choice is very limited with weblog due to its limited functionalities.

Discussion forum on the other hand has more potential to be adopted. It offers almost every characteristics and functionalities we want from our system. There are groups of people, called virtual community, who use both discussion forum and weblog at the same time.

E-mail list Newsgroup W eb forum Chat ro om IM W eblog Cen tralized serv er (online) Y es Y es Y es Y es/No Y es/No Y es Clien t soft w are required Y es/No Y es No Y es/No Y es No v User in teractivit y lev el Medium Medium High High High Lo w T ext supp ort Y es Y es Y es Y es Y es Y es Supp orts group discussion Y es/No Y es/No Y es Y es Y es/No No Async hronous Y es Y es Y es No Y es/No Y es T able 3.1: F unctionalit y and supp ort assessmen t of the applications

3.7

Online Dating Services

A dating system is a specialized meeting system where the objective of the meeting is to go on a live date with someone, usually with romantic impli-cations using any means of technology such as Internet and phone etc. One form of such a system, online or Internet dating service, allows individuals, couples and groups to meet online and possibly develop a social, romantic or sexual relationship. Online dating systems usually archive personal pro-files documenting their members gender, age, build, religion, smoking and drinking habits, self-description, and characteristics preferences in a poten-tial date.

Fiore et al. [9] categorizes online dating systems into search/sort/match (1), personality matching (2) and social network (3) systems. In the search systems, users specify their criteria such as age, interest, and religion, and can search the set of profiles. The matching systems ’pair users by com-paring their profile descriptors to the descriptors of others, usually the con-strained descriptors, because contemporary techniques for clustering or oth-erwise identifying similarity work better with clearly defined features than with free text’. Personality matching systems give personality test to their users and use the results to match their compatibility. Social networks ’en-courage users to bring their friends onto the system and then suggest matches between members of their social networks’ and they offer features such as au-tomatic address book updates, viewable profiles, the ability to form new links through ”introduction services”, and other forms of online social connections. In contrast to other systems that support online communities, online dating systems usually lack a common forum, where everyone can read what everyone contributes and are designed for styles of interaction different from those of the well-known computer-mediated communication environments [2]. However, people can browse through other members’ profiles before deciding to communicate and stay anonymous during the communication. Online dating services allow members from a variety of backgrounds looking for different types of relationships, while some are more specific and may differ in type of members, and their interests, location, or relationship desired.

3.8

Interview Results

The interviews were held in informal environment. Conversations were recorded and saved as audio files and were useful for a follow up. Many useful notions and concepts were suggested and ideas came up during the interviews.

The first interviewee was Mr. Jan Jonson from Avesta. Mr. Jonson was from steel industry and had experience in both planning and attending a conference. He was a member of planning committee of Conference on Thinking 20073_.

He was frequent conference attendee and however did not have previous experience using any support tools. During a conference he usually awaits people to come and start a conversation. The notions from an interview with Mr. Jonson on conference support tool for shared reflection are as follows.

• Classification of lectures and speeches according to their subject and

attendees according to their interest and backgrounds etc.

• In order to classify attendees of a conference, each attendee should have

a profile of keywords of their interested subject.

• These profile subjects should be formalized because informal keywords

and profile setup could be followed by insufficient or misleading search and match results.

• Such a conference support tool should be designed for earlier stage

of communication between conference attendees. The attendees should have possibility to plan their meetings before the conference. People do not have time to use any system during a conference even if the system were superbly designed simply because they would be busy with the conference but not with the system going through all the stages of creating a profile of themselves and browsing through and typing in some texts to search other peoples ideas on the system.

• Accordingly the tool could be used after a conference and become a

communication and interaction platform for conference follow ups and results

The second interviewee was Ivan Rankin from the Department of Sci-ence and technology at Link¨oping University4_{. Mr. Rankin is experienced}

3_{http://www.liu.se/thinkingconference/} 4_{http://www.itn.liu.se/english/}

conference attendee and has computer science background.

There are incidents that list of the people coming to a conference or workshop is not revealed and perhaps even not generated. It would be nice if the system could collect all the information about the people who are to attend the conference. Then one will not have to spend half of the time at the conference looking for someone with same interest or background. Young people with less experience in conferences and public and social events could also benefit from the system.

He also agrees with Mr. Johsson on that there should be a user profile based on keywords and the system should be used for all stages of the con-ference, including earlier, middle and later. He would be willing to use the system during the conference (while he is back at his hotel) if it contains as much information as needed. Other possible uses of the system are for personal purposes. For example, his ”personal contact” wanted to go to an opera and could not get a ticket while attending a conference in Rome. After some vain search on the Internet, he tried some people at the conference and got the ticket instantly.

Chapter 4

Functional Requirements and A

Design Prototype

4.1

Methodology

The second purpose of the project was to develop and recommend functional and non-functional requirements and design choices on the basis of studies in previous step. As starting point for designing a prototype we developed several use case for our system. Use cases manifest how the system should be used and behave. Before developing the use cases we also had to consider our previously developed scenario again and revised it according to our evaluation results. Each use case was illustrated by an activity diagram and its require-ments table. Secondly, we formulated non functional or general requirerequire-ments such as design and usability specifications. Lastly, with functionalities and design specifications and requirements in hand, we materialized the design prototype.

4.2

Functional Requirements and Use Cases

Functional requirements try to capture and express the intended behavior of the system and are described by a use case diagram. The behavior include the services, tasks and functions the system is required to perform. Use case diagrams are based on scenarios. A use case defines a goal-oriented set of interactions between external actors and the system under consideration. Actors are parties outside the system that interact with the system [13].

4.2.1

Scenario

A potential user is about to attend a big conference with more than one hundred participants. There are series of lectures and seminars on various topics. The user will attend lectures that are of interest only to him. After a lecture or speech he would like to meet some people to discuss his reaction, people who might either be working on the same project as him or have the same thought about the topic. A lecture usually involves large number of attendees and every single attendee could have different opinions on the topic. In series of lectures participants don’t have possibility to communicate with every other person and ask their opinions. Therefore he decides to search and initiate communications before attending the conference and meet while he is at the conference.

There is an online system for conference attendees,where registered users can setup their own profile and keywords so that matching is made possible. The systems intention is to help people meet and share their opinions and reactions on lectures or other events that will be (or was) held during the conference and the system is available on the conference web site. The web site contains (possibly) all information on the conference including list of lec-tures and events as well. People have already started sharing their thoughts and opinions well before the conference.

The user goes to the web site and scans through the list of events and lectures. (... ) By logging in he can setup his own profile and choose his keywords, which are essential for matchmaking. The keywords will be generated and assembled by the system administrator or another add-on module using information extraction. In addition to choosing keywords from a previously compiled list, he should be able to put his own keywords. The own keywords could be either unstructured words or a text paragraph. The keyword generating module will use information extraction technique and be able generate keywords directly from lecture notes and publications etc. Afterwards he can search other members by keywords and fields such as name, location etc.

Then user chooses the topic on which he would like to discuss his ideas and, by clicking the link (lecture title), goes to the interaction module. There he reads on lecture notes or abstract and some recent posts to get a general view of other peoples reaction. The posts have nested structure and the user can post a reply message to one of previous message or post a new comment. When the post is submitted he will have list of matching opinions sorted by

their relevance to his opinion. The author of the first match in the list will be considered to have (almost) the same opinion as the poster. If he desires he can decide to initiate a communication with any of the recommended posters or just browse through to read all messages if the number of messages is reasonably small. Every post, including replies because they are considered as a post too, will have a link where their matching posts are displayed. In overview, the user can read posts by other visitors to the site, post a new post or comment/reply to a post, look for matching posts of any single post. The matching module will use the same technique as the keyword generator and, after parsing the text posts, it can match the results.

4.2.2

Use cases

The actors in the Use case diagram (Figure 4.1) are users and the matchmak-ing module. However, the matchmakmatchmak-ing module is not an external subject in the system, but we expressed it as an actor outside the system boundaries. One of the main requirements for the matchmaking module is that it should be possible to embed the module as a component into the system. Moreover, we are planning to prototype the web interface and not the module.

Roles Goal

Submit profile Create a user profile, introduce oneself to other users Post message(s) Express one’s reflection on a topic or a message (reply) Enter search criteria Search for people by their user profile and posted or keyword message(s)

Browse the web site Navigate and acquire knowledge and information Table 4.1: User

Roles Goal

Extract a text (message) Create a set of keywords

Match messages by extracts Find other messages with similar ideas Table 4.2: Matchmaking module

Roles Goal

Display information Present information such as text, user, idea etc. Discussion forum Function as a discussion forum

Steer user Help users navigate through the interface and acquire information

4.2.3

User registration and login

Use case Register/Login

Description The web interface suggests users to register. When user chooses to register himself and submits required information, the system will contact update its database accordingly. Similarly when a registered user submits his identification, the system verifies and authenticates.

Actors User, Web interface (System), Matchmaking module

Assumption Web interface steers the user through the steps before and during the registration process by displaying all the necessary information informing the user his whereabouts etc and responding user inputs. Completeness of the information submitted by the user is handled by the system by prompting before the verification.

Steps 1.User submits his identification 2.The system verifies the identification IF new user THEN

3.Consult the matchmaking module and generate keywords from submitted information

4.Update user profile database ELSE IF incorrect id THEN

5.Reject the user and GOTO the first step 6.User is logged in

Verification The user is logged in successfully.

Variations #1. User might submit new information with intention to update his profile, which will be handled in steps the same as

registering new user.

4.2.4

Browsing and searching user profiles

Use case Browse and search user profiles

Description The user browses through user profiles and he may choose either to read one particular profile or to search by entering a set of keywords. When reading a profile, he decides to contact the owner of the profile or start over again or find other users with similar ideas. When keyword is entered, the system searches user profile database and returns matching users list.

Actors User, Web interface (System)

Assumption User is already logged in. The browsing list and results list are sorted and grouped according to the user’s choice of criteria e.g by registration date and by user location etc.

Steps 1.The user browses users list

IF the user chooses to read a profile THEN 2.Read the profile

IF the user wants to start over THEN 3.GOTO the first step

ELSE IF the user wants to contact the user 4.Initiate a communication and STOP ELSE

5.Search users database and return matching users list and go to the first step.

ELSE IF the user wants to search by entering a keyword THEN 6.Enter keyword

7.Search users database and return matching users list and go to the first step.

Verification The user finds an interested user and initiates a communication. Variations #1. The user might not actually browse through the list,

however the browsing list will still be presented. And the list could be either default list or results list.

#4. The STOP means terminating only browsing/searching activity.

4.2.5

Browsing and searching discussion forum

Use case Browsing, searching and posting on discussion forum

Description The user browses through the discussion forum and can search for posts and can post his own message. Messages can be searched by entered keywords. Every newly posted message or post is extracted by the Matchmaking module. Then when the user searches for messages with similar ideas, these extracts will be compared. The idea of the extraction and how it can be done is explained in Chapter 5. Actors User, Web interface (System), Matchmaking module

Assumption User is already logged in. Writing of comments or replies to previous posts are not distinguished here and will be handled as posting a new message. When displayed the message identification and categorization reference will be the filter.

Steps 1.The user browses discussion forum

IF the user chooses to read a post THEN 2.Read the post IF the user wants to start over THEN 3.GOTO the first step ELSE IF the user wants to see the author’s profile THEN 4.STOP ELSE 5.Search forum database and return matching posts list and go to the first step.

ELSE IF the user wants to search by keyword THEN 6.Enter keyword 7.Search forum database and return matching posts list

and go to the first step.

IF the user posts a message THEN 8.Extract the post 9.Search forum database and return matching posts list and go to the first step.

Verification The user finds an interesting message and starts browsing its author profile (STOP).

Variations #1. The user might not actually browse through the list, however the browsing list will still be presented. And the list could be either default list or results list.

#4. User can choose to terminate the action completely and start another activity such as browsing users list. It is displayed in whole system’s activity diagram (Appendix A)

4.3

General or Non-functional Requirements

Usability/User interface

The system must be easy to use regarding learnability and efficiency. The complexity of the interface should be acceptable for any level of user’s brows-ing and interactbrows-ing skill. The system should not discourage users with its deficient or inconvenient operability.

Availability

The system should be available anytime and anywhere (assuming that user has a web browser)

Share your reflection

The system should support the user in sharing his reflection. The user should be able to express his ideas and thoughts liberally with the help of and by interacting with the system.

Conference support tool

The term conference is used here to support its initial idea and future test implementation and further it would be referred merely as the event. Other alternatives are events and environments such as a TV broadcast and sports games that are widely experienced by public.

Information providing

The system should provide all details on the particular event which it is being used for. For example, users should be able to locate the conference agenda or information on lectures and may be lecture notes. Moreover the user should be informed about the information he has entered and where he is and is going next.

Support for idea matchmaking and dating

The system should help users find their matches by their interests or back-grounds, so that they can share their ideas and interests before the event and consequently reactions and reflections afterwards. Sharing their ideas

before the event will encourage users to meet personally during the event while latter will act as discussion platform. The system should be available online The system should be available on the Internet as a website, so that users would be able to access the system regardless of time and distance.

4.4

Design Prototype

miniBB1 _{was used as a web forum discussion prototype. miniBB (minimalist}

bulletin board) is a flat-type (not threaded) free web forum and discussion software, open source PHP bulletin board, written in PHP and using MySQL or another PHP-compatible database as data back-end. Yet miniBB is easy to use, easy to make change.

Figure 4.5: User registration screen shot

User registration page (Figure 4.5) was changed to accommodate addi-tional fields. Since it is a conference support tool it should be able to register users title, institution, and list of publications etc. Keywords will be gener-ated automatically, still user should be able to enter his own.

Figure 4.6: Admin panel screen shot

Keyword section was added to the Admin panel page (Figure 4.6). Key-words are to be generated by matchmaking module. However, authorized users can create, edit and delete the keywords.

Message Match function (Figure 4.7) will match the message with other people’s messages and give results list of messages sorted by their relevancy. User then can choose to read from the messages and consequently contact and discuss the idea with authors.

Chapter 5

Matchmaking Module and

Natural Language Processing

5.1

Background

The matchmaking module, according to our system architecture, is a com-ponent that extracts excerpts from any text and matches and retrieves these texts. The term text matching usually refers to comparing pair of texts as strings or by their extracted keywords. However, let us assume the following very simple messages that have been posted by three authors who attended same lecture.

The lecture was OK. I liked it.

I am glad I was here.

We can tell that the authors liked the lecture just by looking at them. Can the machine do that?

If we assume that all the input text will be saved in a database, query techniques like key phrase/string matching or indexing can not be efficient enough. However automatic keyword extraction approaches using language model [29] and learning algorithms [30] could be considered in situations dealing with a large amount of texts.

Natural language processing (NLP) is a subfield of both artificial intel-ligence (AI) and linguistics and is sometimes referred to as computational

linguistics. NLP applications are required to have knowledge of language or to understand natural human languages. However, this knowledge may vary depending on the purpose of the application considered. A very simple example is spell checking tool that comes with most document processing applications. The only things this tool needs to know are what is a word (technically) and whether the word exists in its dictionary. A good guide to basics of NLP is a book by Jurafsky and Martin [17] that covers everything related to the field.

5.2

State of the Art

Most of online ”intelligent” matchmaking systems mainly use AI techniques such as intelligent middle-agents and they match services rather than ex-tracted knowledge. For example, Yenta [10] is a distributed multi-agent system that matches people by their interest and taste with the help of in-formation extracted merely from users’ Internet navigation and search history and can arrange possible communication between matched users or agents. While Yenta is being decentralized and based on users web browsing pattern, Ringo [25] is a centralized and e-mail like system where users sign up to the central server and send their own data by rating music they like and dislike. Ringo matches user profiles and then makes music recommendations from matched profile.

Nasukawa and Nagano [21] have developed TAKMI (Text Analysis and Knowledge MIning), a system which finds valuable patterns and rules in a text, the rules that indicate trends and significant features about specific topics when applied to large textual data. The system makes use of intention analysis and dependency analysis to extract concepts or text representations from given data. Statistical analysis functions were applied afterwards on these concepts and results could be analyzed interactively with graphics. Richardson [22] identifies similarity patterns of semantics relations between words, that are already tagged by natural language parser for a dictionary, by assigning weights to the relation.

5.3

Information Retrieval (IR)

Information retrieval (IR) is a branch of NLP that retrieves documents and other types of data specified by a query within a database, Internet or a domain. Document retrieval (DR) and text retrieval (TR) are different prac-tices with different theories and IR is a global term that covers both and others. While DR retrieves (online and offline) documents, such as articles and journals, TR supplies users with end text.

Text REtrieval Conference1 _{(TREC), part of the TIPSTER text}

pro-gram2_{, is the leading community that experiments information retrieval}

sys-tems in different domains with different aims to provide the infrastructure for evaluation of text retrieval methodologies. IR task is ”accomplished using statistical methods that (a) select terms (words, phrases, and other units) from documents that are deemed to best represent their content, and (b) create an inverted index file that provides an easy access to documents con-taining these terms” [26].

5.4

Information Extraction (IE)

Information Extraction (IE), also a branch of NLP, is a technology that analyzes and transforms natural language text into structured format and thereby reducing the information in the document to a tabular structure in order to extract snippets of information [8, 14]. The process takes texts as input and produces fixed-format, unambiguous data as output.

An example from [1], which looks very similar to our case study and could explain some motivations of pursuing IE is ”an IE system designed to monitor technical articles about Information Science [that] could pull out the names of professors, research studies, topics of interest, conferences, forthcoming publications from press releases, news stories, or e-mails and encode this in-formation in a database. End-users can then search across this database by textual attributes or features. A typical search could be for all forthcoming publications about information retrieval or to locate all conference presen-tations on a specific information science topic. In addition, the structured information contained within a database could be ordered into a taxonomy”.

1_{http://trec.nist.gov/}

The difference of IE from Information Retrieval (IR) is that IR system finds relevant texts and presents them to the user as a whole (documents), while IE system analyzes texts and extracts only the specific information, or the facts, that the user is interested in. Therefore IE systems require predefined queries as to what kind of information the user wants extracted from a text. While the documents returned by information retrieval systems have to be read and analyzed by humans, the database entries returned by information extraction systems can be processed by data processing programs [14].

Information extraction involves tasks such as entities recognition (per-sons, locations, time ...), coreference recognition, establishing relationships between entities and events and domain specific inference. Domain specific IE systems have better performance due to accuracy in results returned and ease of development. For more on IE, start with AI TOPICS3 _{by the}

Amer-ican Association for Artificial Intelligence4_.

5.5

Performance Level

Several research projects, sponsored by U.S. Navy, were on track of extract-ing naval messages in the late 1980s and led to a new research field, IE. Later on, Message Understanding Conference (MUC) was started to eval-uate IE systems and compare their performance. Human performance was considered as a point of comparison when evaluating IE systems test results. Participating IE systems are evaluated by completing defined tasks on test sets of text documents and extracted information is compared and scored against information manually extracted by human analysts.

The Automatic Content Extraction program5_{, a successor to MUC, has}

been running since 1999. The ACE tasks are more complex than their MUC counterparts and evaluation results are not public.

Evaluation metrics of IE, recall and precision, were adopted from IR met-rics. Recall measures how much information was extracted, while precision measures how much information extracted was correct. Both metrics are on the interval [0, 1] and inversely related to each other. Both should be con-sidered when comparing IE systems. However it is not straightforward to

3_{http://www.aaai.org/AITopics/html/info.html} 4_{http://www.aaai.org}

compare the two parameters at the same time and, as a roundabout, various combination methods have been proposed, such as F-measure [8].

5.6

IE Tasks

IE is split into five tasks [19]:

• Named Entity recognition (NE) finds and classifies person names, places

etc. Multi-Lingual Entity recognition (ME) is optional variant for Chi-nese and JapaChi-nese language set.

• Coreference resolution (CO) identifies identity relations between

enti-ties.

• Template Element construction (TE) adds descriptive information to

NE results (using CO).

• Template Relation construction (TR) finds relations between TE

enti-ties.

• Scenario Template production (ST) fits TE and TR results into

speci-fied event scenarios. Consider these sentences:

Telephone providers Verizon Communications and AT&T Inc. on Thursday posted financial result. Stifel Nicolaus analyst Chris King said it showed promising growth.6

NE discovers that the entities present are the Verizon Communications, AT&T Inc., Thursday, result, Stifel Nicolaus, Chris King, growth. CO discov-ers that it refdiscov-ers to the result. TE discovdiscov-ers that the Verizon Communications and AT&T Inc. are telephone providers, result are financial results and the growth was promising. TR discovers that Chris King works for Stifel Nico-laus. ST discovers that there was a financial result posted by the companies and Chris King said something about the result.

ACE combines these 5 tasks into only three tasks. The NE and CO tasks become a single Entity Detection and Tracking (EDT) task in ACE, and the TE and TR tasks a single Relation Detection and Tracking task. The ST task is renamed Event Detection and Characterization.

5.7

IE Systems

The following paragraph is excerpted from a tutorial by Appelt and Isreal [3], which demonstrates IE systems and its debate.

There are two basic approaches to the design of IE systems, which are labeled as the Knowledge Engineering Approach and the Automatic Train-ing Approach. In knowledge engineerTrain-ing approach, information parsTrain-ing and extracting rules are defined by knowledge engineer or domain expert. The expert will formulate his rules based on pre-defined and usually domain spe-cific text and his own knowledge in the domain. Despite its disadvantage that it is highly dependent on the knowledge and expertise of the engineer, most of the well performing IE systems are based on this approach. On the other hand, the automatic training approach does not require human knowledge but labor. The system will be fed by training corpora, a set of domain relevant texts, to acquire grammar annotations and then a learning algorithm is run resulting in information needed for further analysis.

OpenNLP7 _{is an online list of open source projects in NLP including IE}

and IR, available for both scientists and developers. For example, Minor-Third [6], a collection of open source Java classes, has a number of learning methods that extracts text and label text documents. The most popular of IE projects is General Architecture for Text Engineering8 _{(GATE), which offers}

not only classes and components but also a software development architec-ture and component development tools. The Natural Language Processing Research Group at the University of Sheffield, UK9 _{and Information Sciences}

Institute10 _{have also bunch of links to IE projects.}

7_{http://opennlp.sourceforge.net} 8_{http://gate.ac.uk}

9_{http://nlp.shef.ac.uk/research/areas/ie.html} 10_{http://www.isi.edu/info-agents/RISE/projects.html}

Chapter 6

Conclusion

We have reviewed several types of mass interaction systems and assessed their similarity to our desired system. With the rapid development of the Internet, computer mediated communication is evolving more rapidly. From evaluated applications, web discussion forum turns out to be the most versatile method of sharing reflection during an event. Potential conference organizers and at-tendees were interviewed and gave their view on the conference support tool, shared reflection and shared their own experiences. Based on evaluation re-sults, system functional and non-functional requirements were structured and design prototype was outlined. Finally, the matchmaking module was dis-cussed and Natural Language Processing and its subbranches were suggested for the implementation.

Bibliography

[1] K. C. Adams, The web as database: New extraction technologies and

content management, Online (2001).

[2] Jeana Frost Andrew Fiore and Judith S. Donath, Scientists, designers

seek same for good conversation: Workshop on online dating.

[3] D. E. Appelt and D. J. Israel, Introduction to information extraction

technology, 16th International Joint Conference on Artificial Intelligence,

1999.

[4] Susan B. Barnes, Computer-mediated communication: Human-to-human communication across the internet, Allyn and Bacon, 2003.

[5] Hee-Kyung Cho, Murray Turoff, and Starr Roxanne Hiltz, The impacts

of delphi communication structure on small and medium sized asyn-chronous groups: Preliminary results.

[6] W.W. Cohen, Minorthird: Methods for identifying names and

ontologi-cal relations in text using heuristics for inducing regularities from data,

2004.

[7] Pavel Curtis and David A. Nichols, MUDs grow up: Social virtual reality

in the real world, COMPCON, 1994, pp. 193–200.

[8] Line Eikvil, Information extraction from world wide web - a survey, Tech. Report 945, Norweigan Computing Center, 1999.

[9] Andrew Fiore, Online personals: An overview.