ARTIFICIAL CONVIVIALITY

Inventing Communities of Communication

In the physical world, humans interacting together are remarkably skilled at using subtle cues about their presence and activities. However, in digital worlds such interactions are still limited. While analyzing these limitations we will aim at defining a conceptual framework as well as requirements to allow for more effective online interactions.

We will first look at how humans interact in the real world: Interactions can be between single individuals, an individual and his/her environment or individuals and groups with their environments. We will ask what types of inputs and outputs could be appropriate for such online interactions? What kind of mechanisms, functions, roles and rules should be defined to regulate these interactions? What properties should such environments be characterized with?
Our research is embedded in web site environments such as digital cities, online-portals from city administrations to service their citizens.
We will exploit the graph-like structure of web sites, where pages are nodes and links edges, as spatial environment and stage for interactions between individuals, the visitors of the site, alone or in groups and the site itself.
Furthermore, we will consider the notion of conviviality and how it relates to interactions between individuals as well as between environment and individuals.

Setting up the Problem

We do not have in cognitive and social science a clear notion of conviviality. There is no clear model nor any singular vision of conviviality. The definitions we have found are either incomplete or domain specific and we will here attempt to define it.
Simply stated, a convivial place is a place where one is welcome, one feels at ease. However, we believe that conviviality is a concept of greater depth that plays an important role not only in social interactions but also in the internal regulation of social systems and that it deserves closer scrutiny.
A convivial relation between individuals is one that allows participating individuals to behave and interact with each other following a set of conventions either shared, commonly agreed upon or at least understood. This presupposes implicit or explicit regulation mechanisms based on concensus or "social contracts" and applied to the behaviors and interactions of participating individuals.

Related concepts: amicable relation, politeness.

A convivial relation is to be distinguised from an intimate or amicable relation. Similarly to politeness a convivial relation takes place between individuals that do not necessarily know each other and requires that a certain distance be respected between the individuals. We plan to analyze this distance and attempt to evaluate it.
We see conviviality differing from politeness in that conviviality adds the notion of "well-being for the group" to the set of shared conventions whereas politeness may keep relation only effective and without direct confrontations between single individuals (ex...).

Discussion

Let's look at the notion of environment as described by Weyns et AL. [1], e.g. a spatial and organizational structure relating to activities such as communication and the definition and enforcement of rules regulating the interactions within it.

How do individuals within a group interact with each other? What are the functions and roles played by individuals while interacting?
Is conviviality a tool to strike a balance in the group between the benefits for individuals alone and benefits for the group at large?
What is a convivial community? How does it form, dissipate, evolve and reform?

How to describe, evaluate and model ''conviviality'' in digital cities

Let's first define three functions.
The domain of our function is a group of individuals (for example visiting the digital city web site.)

In (1) conviviality is described as a function over all users. The function computes the conviviality of the group. The function parameters are the conviviality evaluations of each and every user composing the group.
In (2) we obtain for each user, his/her own conviviality evaluation. The function parameters belong to a set of conviviality criteria. These criteria could for example be pre-determined for the group or dynamically generated by users. Examples of criteria: trust, quality of interactions, etc.
In (3) each criterion is computed over the set of parameters defining that particular criterion. For example: the parameters of trust could be the security level of the environment, information about users, etc. We plan to analyze criteria in more details.

Example of Scenario:

On the digital city site, a user looks for tourist information about museums. Navigating through the site, a search box or a companion agent, s/he arrives on the museum page. There are already a number of users. How can conviviality be expressed, assessed and managed?

References

[1] Weyns, D., Parunak,V.,Michel, F.,Holvoet, T.,&Ferber, J. (2005c). Environments formultiagent systems, state-of-the-art and research challenges. In Weyns et al. (2005a). Springer-Verlag.

I2 - INVENTING PROFILE DISCOVERY FOR E-CONVIVIALITY

Conviviality in digital cities is per se rather known, less explored and implemented. Following the definition of a natural conviviality, we also have observed a very interdisciplinary way of interpretation and definition, both in open and closed social systems. Following this, the intended research work can be only a fundamental research work that finally leads to the implementation of prototypes. Starting from conviviality in a natural social system - and especially how it is expressed - the central questions of the screenplay might be:?

Natural conviviality

What is natural conviviality? Which advantages does it have, e.g. from a evolutionary point of view? What is the difference to comfortability, usefulness, etc.?
Which parameters are currently seen as to be responsible for a natural conviviality? How are they recognized, expressed, and represented (managed)?
Are there differences in an open and closed (= only persons who know each other) convivial system? Do their value change over time, i.e., are parameters dynamic with different importance?
In this respect, what is the role of trust, i.e., does it play a central role for natural conviviality? As for natural conviviality, how is it recognized, expressed, and managed?
What is the role of knowing the user in a natural environment? Do humans tend to feel more comfortable and convivial if they know who other users behave?
Is an natural conviviality time- and user-dependent, i.e. what happens if a person joins or leaves a group of human individuals??

Artificial conviviality

What means artificial conviviality in a digital city? There is probably a need for a definition ([which is probably the result of the PhD work])
Is artificial conviviality split into sub-disciplines? If yes, which ones? Do there exist research material?
Following current technologies, what means digital conviviality in a hybrid scenario (humans, artificial entities/actors)?
Do there exist already research work and/or implementations? Overall, how was the implement of a digital conviviality be performed?
Which kind of implementations may be of sense (think big - start small)?
How useful is it at all to realize artificial conviviality? Will it have significant add-on value for the City of Luxembourg? ([must be answered in the end])?

Pattern Discovery

Assume, we express a digital conviviality through a conviviality function having different parameters as arguments.? * Are the parameters of a natural conviviality to be transferred to a artificial conviviality one-to-one or do other (which do not occur in the natural scenario) parameters also (or even stronger) contribute to the digital conviviality?
How can we measure a common conviviality, i.e. how is the conviviality function evaluated? What do we do with the value?
Are the parameters absolute or relative, i.e., are the parameters different from user to user? In this respect, can we find groups of users and therefore reducing the number of individuals to a few?
Certain hypotheses:
- Hypothesis 1: the parameters of the conviviality function are completely fluent/dynamic: at a certain time it is not known which parameters are active and which not. Furthermore, it is not known which parameters play a higher role than others.
- Hypothesis 2: the parameters of the conviviality function is completely static. At a certain time, we always have a fixed list of parameters.
- Hypothesis 3: the parameters of the conviviality function are sometimes dynamic and sometimes static. This is known.
- Hypothesis 4: the parameters of the conviviality function are sometimes dynamic and sometimes static. But this is not known (see then Hypothesis 1).

General questions, depending on the hypotheses:

How fluent/dynamic are which parameters? Do we then need to concentrate on all these parameters?
May dynamic parameters become static and vice versa?
How can we then discover the active state of parameters at all?
How can we measure their importance (weight)?
Can we apply these parameters to all users or are they individual?
Do we have to distinguish in a hybrid scenario?
It may be that several hypotheses are true under certain conditions (time, hybrid scenario, etc.)?

In this respect, additional things should be examined:

Time windowing: to which extend must the pattern (parameter) discovery adjusted (day by day, month by month?)
User transparency: to which extend do we need profiles of the (human) users? If yes, how to use them into the conviviality scenario and when to compute them? We may assume that they are static, but in principle they are dynamic.
Do we have to take into account all users or only these who significantly contribute to conviviality?
Do we need the user’s role to ensure artificial conviviality? Which roles might be appropriate? Are roles fluent and when do they change?

Implementation

How can we realize such a platform? In the beginning, we may receive some experiences from setting up a simple scenario with several users and measuring their quality of life while improving this continuously (following several dimensions).
How can we initialize a dynamic conviviality function? How and when do we update the conviviality function? How do we manage it, for example in a “conviviality center” or more individually?
How can we realize a learning environment where changing parameters (weight, grade of dynamics) can be discovered? Is there some trigger mechanism saying that a parameter may change?
How to realize a common framework with hierarchies (individuals, groups).?

Goal of this sub-project

Finding a definition of what conviviality is in digital city, where it depends on, etc.
Validating and proving the worth of conviviality at all.
Realization of a discovery and management platform as described above
Having prototypical implementations that show the conviviality under the circumstances of the definition (at the City of Luxembourg).
The candidate’s opinion on what the advantages and disadvantages are, if external influences (for example the re-adjustment of parameters of artificial entities) should be allowed, how performing the implementation is, what future aspect will be, etc.

Inventing profile discovery for e-conviviality

The core assumption of this work is: a visitor feels more comfortable if he gets the desired information easy, fast and smooth. To achieve e-conviviality means to know the visitor’s desires (user profile) and to provide him with the information he needs and/or may need. The approach of this work is to solve this problem by using profiles, namely profiles of behavior and profiles of interests. These profiles will be built and analyzed. The results will be used to increase and improve a user’s e-conviviality in digital cities e.g. show the desired content...

f

Staff

Results

"ICC" is mentioned on: MINE Projects