Mistica | homepage

Virtual Community

Your previous page Project Overview | Pilot Projects | Clearinghouse | Events | Evaluations | Cyberlibrary Your next page
[ VC Home Page ]
[ Subscribe/Unsuscribe VC ]
[ VC Rules ]
[ VC Members ]
[ EMEC Support Documents ]
[ VC Production ]

The EMEC Methodology

(Efficient Management of Multilingual Electronic Conferences)

Knowledge Management in a Latin American Virtual Community


Daniel Pimienta, Catherine Dhaussy

November 2001



Since the mid-1990's, shifts of the Internet towards business and entertainment have called for innovative approaches in the management of virtual communities. Fashionable web conferences often fail in communication aspects; the information they provide, although nicely organized, cannot replace computer mediated communication. Chat tools, on the other hand, suffer from the opposite flaw: synchronous communication is generally poor at good information management. This paper claims for an alternative approach focusing more on methodology than software to generate communication and to disseminate information. The methodology aims at decreasing the level of the barriers (information overload, time, language…), hence increasing the incentive for proactivity. The experiment allowed to facilitate the management and to improve the productivity, through a database system automating routines. EMEC, the proposed methodology, allows a better thematical focus, and it respects people's schedules, languages, and cultures. Designed in 1997, it was tested in 1999 as a research-action product, in the framework of a Latin American virtual community dealing with social impact of the Internet. This study presents the first results of the experiment, and opens perspectives for future applications. The next version of the methodology will focus on customization, in order to increase users satisfaction, and will open a new field of research targeting "facilitation of distance participation".

Keywords: virtual community, knowledge management, methodology, machine translation, Latin America


1 Introduction

The mass marketing of the Internet has increased the number of people using e-mail services, a trend which has logically led to a growth of mailing lists, both in variety and size. The majority of the new users have not followed the acculturation process which was the norm until the mid-90's. This has generated an increase of inappropriate messages (especially in content, and length) in e-conferences. Moreover, the greater number of lists with similar objectives makes users more exposed to cross-postings. Additionally, the shifts of the Internet towards commercial media have increased the phenomenon of spamming; furthermore, virus spread by e-mail is now generalized and does affect e-conferences. Finally, the internationalization of the Internet drives to a more diverse audience in discussion lists; it has thus become a priority issue to accommodate non-English speakers and crosscultural exchanges.

The combination of those factors has produced a less and less acceptable amount of wasted time for professional subscribers. The dramatic reduction of the density of useful information leads to a drop of the attention devoted to the main messages, which get diluted in the incoming flow; hence a threat to the efficiency of e-conferences. Experts leave thematic conferences, which have been invaded by a large poorly qualified audience; this phenomenon contributes to a qualitative decrease as far as information is concerned. In a few words, the global signal/noise ratio of e-conferences has gotten down to dangerous thresholds. Despite these issues, e-conferences should remain the main basic principle of Internet-based collaborative work; some method is then required to adapt them to the new environment.

The fast growing use of online conferences (web-based forums) has not solved the problems linked to adequate participation. Those systems often offer an impressive interface, they may allow a good information management structure but they lack the communication incentive which makes the soul of virtual communities.

Professional moderation of closed discussion lists is obviously part of the solution, but this is not enough. In order to face the set of identified problems, FUNREDES, an NGO dealing with the dissemination of the Internet in the South, has designed a methodology called EMEC — "Efficient Management of Multilingual Electronic Conferences". The process, applied on the top of moderated distribution lists, was tested for one year (starting mid-99) in the framework of the MISTICA project — the Spanish acronym for "Methodology and Social Impact of Information and Communication Technologies in America".

In this paper, after having presented the methodology, we shall continue with a description of the experiment; then, we will analyze the first results, and will conclude with some perspectives for the future.


2 The EMEC Methodology

EMEC deals with the creation of a moderated electronic conference system based on list servers with added value elements aiming at removing currently existing inconveniences which restrict participation; additionally, machine translation is incorporated into the process, to open the space for crosscultural communication.

The set of multilingual services which are implemented aim at sustaining the communication flow of the e-conference, and reducing the amount of unsolicited incoming information while providing easy retrieval of the original contributions. The central elements are:

  1. the creation of contribution headers with an information management orientation, the offering of an abstract of the original contribution, and the organized archiving and retrieval of the contributions in various languages;
  2. a permanent focus on users training, both implicitly and explicitly.

As far as basic principles are concerned, EMEC combines classic e-mail with Web resources. Four main axes may be identified at the operative level:

  1. subscribers receive the contributions in the form of brief messages with the key data of the message and a summary of its content, along with a web reference linking to the page where the original message is available;
  2. the original messages are classified and uploaded on a web page;
  3. both the presentation note and the integral message are translated into the supported languages;
  4. the contributions are automatically archived and retrieval is possible, either from log files (Hypermail, e.g.) or by queries to a specially designed search application.

Users can enjoy several innovations:

  • on the level of incoming mail: a very brief informative note instead of a whole message, reading of the integral message on the Web, use of a selected language;
  • on the level of outgoing mail: use of a language according to personal linguistic preferences and abilities, requirement to respect given writing rules in order to make the automatic translation program work efficiently;
  • on the archives level: classic e-mail logs; additionally, structured archiving for the original messages;
  • on the community charter and on the level of individual commitment: new rights (read and write in one's own language, have one's message diffused without having been translated, have a minimum load of messages, save time in the administration of one's e-mail, have well organized files on the Web) as well as new duties (get used to the context, respect writing rules, go to the Web to read integral messages).

It has become quite common to mix e-mail push and web-based pull in the management of virtual communities. However the EMEC methodology articulates the two supports in an original way. The message sent through a list server to all the members of the virtual community contains three main parts:

  1. raw information about the message: author, title (may be edited during moderation; in the case of a multi-thematic e-conference, the first word indicates the topic), date and time sent, language of the original message, size of the original message or of the translated original message, keywords;
  2. a synthesis of the original message;
  3. the URL of the original message or its translation.

The website is divided into four parts:

  1. miscellaneous information about the given electronic conference and virtual community, including explicit presentation of netiquette rules and a charter on communication in virtual forums;
  2. information about the EMEC methodology;
  3. the list archives: both the messages sent to the list and the integral texts of the postings are on the Web; the files transmitted as attachments are organized in a digital library;
  4. a search area, which offers the possibility to perform an advanced multi-criteria search.

The following links show an example of one EMEC-processed message:

For both time and budget reasons, it is almost always impossible to have human translators throughout the process; the use of a computer-assisted translation program is required. EMEC does not use only automatic translation, however. As far as the key data and abstract are concerned, the translation is entirely made by a human team. The situation differs for the whole text: the message is then pre-processed, i.e. it is roughly re-written according to a set of rules allowing the program to perform better, hence to give an understandable result. People are then encouraged to see the automatic translation merely as a tool which may help them to have a broad idea of the content. Moreover, anyone with a basic knowledge of the original language is invited to use the original message as a tool able to clarify the automatically translated text.

The EMEC methodology is supposed to be transparent for its users: the main part of the load is supported by professionals, which allows the members of the e-conference to concentrate on the very content, and to stop bothering about what does not belong to content itself. However, a quite heavy logistics supports the effort and there is obviously an associated cost per message.


3 The Experiment in MISTICA

The chance to experiment EMEC was provided by IDRC (International Development Research Center of Canada) in the framework of the project MISTICA, launched late 1998.

EMEC was introduced in September 1999. A regular moderated e-conference in Spanish had been active for six months, among a virtual community of some 180 subscribers. The original mailing list was divided into four lists, run in parallel. The content was strictly identical, the only difference lying in the language: Spanish, Portuguese, French, and English were supported, i.e. the main four languages of the area.

Members of the virtual community had been prepared to the shift for two months: step by step, weekly messages made them aware of the new methodology which was about to be implemented. However, these messages did not receive much feedback, even when people were asked to chose the language in which they wanted to read their incoming e-mails.

A little less than 400 EMEC-formatted notes were sent to each list during the experiment, i.e. an average of 1 to 2 per working day (up to 6 a day). Many other messages were received by the moderation team, and rejected because they had not followed the rules (spam, out of focus, administrative matters...). For transparency reasons, out of focus messages were temporarily stored in one of three archive categories, the other two being valid messages, and messages limited in time (e.g. conference announcements).

As far as processing cost is concerned, the figure was around US$ 20 per contribution (for 4 languages and with salaries corresponding to an NGO working in the South) and this amount lowered to US$ 10 when the process was automated, at a later stage of the experiment.

The EMEC team was composed of 4 people, a supervisor, and the project manager. The moderation was made by a pool including the Mistica coordination group and the EMEC team supervisor; the moderation was considered too touchy a skill to be delegated. The EMEC team included a full time person responsible for the conduction of the whole process, including the writing of the syntheses (the second sensitive part of the process) and machine translations, and two persons responsible for translating the syntheses and updating the website. The last member of the team was responsible for the design of the automation process (using PosgreSQL under PHP/Linux,) having the working interface designed in close collaboration with the rest of the team. The EMEC program allowed to double the productivity of the team by facilitating and integrating the input/output process through a single interface to the Funredes Linux server. The member of the teams were young European professionals recruited by Funredes via the Internet, willing to share a two year experience in a developing country and holding a skill of librarian with fluency in 3 languages.

The EMEC team learnt many lessons in one year. Most of our hypotheses were confirmed by the experiment, as described in the next part of the paper.


4 The Evaluation and Lessons Learned

The experiment was evaluated by the subscribers in June 2000. This questionnaire received more responses than the previous evaluations of the MISTICA project. Many subscribers who had never sent a contribution to the list participated and showed their interest with the experiment.

We can classify the population of the virtual community into:

  • proactive (less than 10% of the total of subscribers),
  • active (less than 30%),
  • passive subscribers (more than 60%)

According to this scheme, the trend was:

  • the proactive group was quite upset by EMEC : they feel it lowered their spontaneity and proactivity;
  • the active group saw both advantages and disadvantages (the majority would not stand the poor level of machine translation, many were irritated by a third party making a short abstract of their contributions, most of them acknowledged the benefits of maintaining focus and size);
  • the silent majority (passive group) expressed their high satisfaction with a device allowing them to stay aware with a minimal time investment.

A provocative way to express part of the outcomes is that some subscribers were happy to see other subscribers messages processed through the EMEC fashion but quite unhappy when their own messages were processed that way, as they see it as an offence to their production... The high level of expectation with machine translation from the subscribers who had not paid attention to the recurrent warnings from the EMEC team about the obvious limitations of that service was a difficult issue to face. Frequent complaints about the unacceptable quality of translation were made. The team wanted to use another term than "translation" to avoid the misconception, something expressing "a minimal product to help the understanding for non-speakers", but no agreement was reached by the end of the project.

Proactive subscribers felt that the processing time, much higher than in the case of a basic moderation (up to a couple of days, vs. one hour), set too big an obstacle to spontaneous and lively exchanges. Most proactive Internet users have an adequate information management organization: they may also have felt useless a device reinforcing external structure while their own decentralized structure already answered their needs.

As a matter of fact, the unsubscribe rate in MISTICA has always been much below standards and the civil society context of this virtual community allowed a democratic escape of the tensions created (meta discussions on the management of the virtual community and the EMEC processing were accepted in the conference and final decisions were made by "rough consensus").

Other experiments need of course to be carried on, with other groups, in other frameworks, to reach generalizations on the evaluations and results. The results could be different with a distinct virtual community; our experiment was made within a group characterized by a large majority of Spanish-speaking and ICT professionally skilled subscribers.

As far as knowledge management is concerned, EMEC gave general satisfaction to the group: the methodology allows to save time, to minimize the information overload, and to reduce the communication gap between people from distinct cultures and backgrounds. In this sense, it fully reached its original objectives.


5 Perspectives

The experiments conducted to design appropriate methodology for multilingual virtual communities started a long time ago as a will to create an environment of fluid and efficient performance for online groups from various backgrounds, an environment which would protect the members from information overload, and help to develop the proactivity within the community. Its documented and proven methodology, which has built a reliable framework while allowing some flexibility, makes EMEC replicable in many collaborative environments, whatever the topics and linguistic background might be.

Proactivity determined a whole range of satisfaction: hence, it appears that a methodology able to customize the settings would be likely to satisfy all participants. This is the essence of the second version of EMEC: each member of the virtual community would define the form of interface with the list that (s)he prefers, as far as synthesis and translation are concerned.

Another line of improvements concerns the scope of languages: indigenous languages (Quetchua, e.g.), Creoles (Haitian, e.g.), should be added to the list of languages supported in the future, as long as corresponding machine translation software get available.

Finally, Funredes will open a new line of research in incoming projects (like in http://funredes.org/olistica ), which consists of adding a new and independent layer on the top of EMEC, aiming at enhancing participation in virtual environments. The basic idea is that efficient face-to-face meetings are chaired by highly-skilled professionals, often called "facilitators", which help people to participate evenly, balance creativity and focus, help to solve group conflicts and, as a general rule, facilitate the communication flow in the conference room. The concept would be then to recreate this profile and the corresponding tools in an environment which is deeply different in many communication patterns (asynchronous, no physical feedback, written expression...).



6 Useful Links – Further Readings


7 Authors

Daniel Pimienta < [email protected] >: Head of the Networks & Development Foundation ( http://funredes.org ) since 1988, he had previously worked for 12 years as a System Architect and Planner (IBM France and USA). Settled in Dominican Republic for almost 15 years, he has since then devoted his work to helping development through the use of ICT. He designed EMEC as an original project as early as 1997. Pimienta studied Applied Mathematics and holds a Ph.D. on Computer Science from Nice University in France.

Catherine Dhaussy < [email protected] >: After some projects linked to virtual communities, and a graduate research on the topic, she joined Funredes late 1997. When the EMEC project was launched, she became responsible for the coordination. Presently living in France, she is completing a Ph.D. in crosscultural studies.

Your previous page Top of the page Your page suivante
Search | Credits | Site Map | E-mail: <[email protected]>
Last update: 25/04/2000