=Paper=
{{Paper
|id=Vol-2134/paper08
|storemode=property
|title=Modelling the Social Environment: Towards Socially Adaptive Electronic Partners
|pdfUrl=https://ceur-ws.org/Vol-2134/paper08.pdf
|volume=Vol-2134
|authors=Ilir Kola,Catholijn M. Jonker,M. Birna van Riemsdijk
|dblpUrl=https://dblp.org/rec/conf/ijcai/KolaJR18
}}
==Modelling the Social Environment: Towards Socially Adaptive Electronic Partners==
https://ceur-ws.org/Vol-2134/paper08.pdf
Tenth International Workshop Modelling and Reasoning in Context (MRC) – 13.07.2018 – Stockholm, Sweden
Modelling the Social Environment:
Towards Socially Adaptive Electronic Partners
Ilir Kola1 , Catholijn M. Jonker1,2 , M. Birna van Riemsdijk1
1
Interactive Intelligence Group, TU Delft, The Netherlands
{i.kola, c.m.jonker, m.b.vanriemsdijk}@tudelft.nl
2
Leiden Institute of Advanced Computer Science, Leiden, The Netherlands
c.m.jonker@liacs.leidenuniv.nl
Abstract terms have been used to describe attempts to enable artificial
agents to better understand their surrounding environment.
Providing support to a person to manage the activ- According to Barwise [1987] these concepts refer to the same
ities of daily life requires a framework to represent thing, and situations represent a way of modelling contexts.
the social environment of that person. The frame- Situation awareness is primarily used to model emergency sit-
work introduced in this paper allows the modelling uations such as rescue operations, therefore the main focus
of the subjective experience of that person in spe- has been on modelling the physical environment. However,
cific situations, the social relationships of people Kaminka [2013] argues that agent systems should incorpo-
playing a role in that situation, as well as general rate general social intelligence building blocks. Dignum et
knowledge that can be used to derive additional in- al. [2014] also suggest that the next step in artificial intelli-
formation about specific situations. gence is the ability to show social intelligent behaviour.
In this paper we focus on modelling situations for Socially
1 Introduction Adaptive Electronic Partners [Van Riemsdijk et al., 2015],
which are artificial agents that support people in their daily
Artificial agents that support people in various activities are lives, and base this support on the users’ preferences, i.e., the
becoming a reality, consider e.g., virtual coaches [Tielman et decision when to support and the form of support is tailored to
al., 2014], personal assistant agents [Myers and Yorke-Smith, the user. We use the Situation Theory Ontology (STO) frame-
2007], and smart homes1 . Simon [1972] proposes that human work developed by Kokar et al. [2009], and we extend it in
behaviour is shaped by the computational capabilities of the order to enable it to account for social relations. We choose
actor as well as by the structure of the task environment. This this framework since it allows the representation of arbitrary
suggests that in order for artificial agents to be able to support situations from different points of view. The resulting knowl-
people in their daily lives, information about their internal edge structure enables us to model and reason about situations
processes is not enough: it is also important to represent the in which the social component plays an important role. In
environment. In simple words, help should be situational and Section 2 we introduce a motivating example. The necessary
therefore, we need to know what is going on around us. To background knowledge used in our approach is discussed in
provide assistance that fits the situation, an agent should be Section 3. We present our approach and use it to formalize
able to reason about the surrounding entities and how they the example scenario in Section 4. We wrap up the paper and
relate to each other. discuss future work in Section 5.
If we turn to sociology, the concept of definition of the sit-
uation is considered to be what people use in order to know 2 Motivating Example
what is expected of them in a given situation. It is a subjec-
tive understanding of the role and status of those involved in In order to illustrate our approach we present an example sce-
a situation. We learn how to define situations by combining nario which will serve as the base for the framework in Sec-
our experiences with our knowledge of norms, customs, be- tion 4. To make the scenario appropriate for our goals of sup-
liefs, and social expectations. The term first appeared in Park porting people in their daily life, we include settings in which
and Burgess [1921], who write: “...In fact, every single act, social relations influence the actions that are to be taken.
and eventually all moral life, is dependent upon the definition In the scenario our main character, Alice, and her two
of the situation. A definition of the situation precedes and friends, Bob and Charlie, are going on a holiday trip. The
limits any possible action, and a redefinition of the situation first step is for them to decide when and where to go. After
changes the character of the action.” deciding, Alice notices that on the morning of departure she
Research in computer science has tackled this issue by us- has an important meeting with her boss at work. During the
ing concepts such as situation awareness [see Endsley, 2000] holidays while on a picnic, Alice forgets to bring something
and context awareness [see Akman and Surav, 1996]. Both to drink, so she asks Bob for a beer. Later that day when
they go to a shop, Alice buys a chocolate, and then gives the
1
http://sine.ni.com/cs/app/doc/p/id/cs-14844 chocolate to Bob.
30
Tenth International Workshop Modelling and Reasoning in Context (MRC) – 13.07.2018 – Stockholm, Sweden
We want to model the situations in this scenario from the under that specific situation. In this formalism, s φ means
point of view of our main character. For this reason we need that the infon φ holds in situation s. Since situations are rep-
a knowledge structure based on a subjective representation resented formally, it is possible to infer new facts by using
of the world. The knowledge structure should explicitly rep- OWL formalisms. Furthermore, it is possible to construct
resent different social relations between people, taking into more general rules and then infer information by combining
account that people can have different relations in different them with the facts present in the ontology.
situations. Furthermore, we need the possibility of combining Many of the concepts presented so far are similar to our re-
the facts from the situation with more general rules in order quirements: situations are seen as a partial subjective descrip-
to infer new information about the situation and possible ac- tion of the world from the point of view of a user, it is possi-
tions. Lastly, it would be useful to have access to information ble to refer to other situations, the framework is generic and
from past situations during the reasoning process. expressive, and it is possible to add rules, which in our case
can represent societal norms. Furthermore, the formalizations
3 Preliminaries can be graphically represented, which gives the possibility of
The approach introduced in this paper is based on existing interactively building a model of the situation with the user.
concepts from situation awareness and sociology. In this sec- For other approaches used to model context, see Akman and
tion we present these concepts and discuss why they were Surav [1996]. However, at this point the STO framework does
chosen. Due to space restrictions, our description does not not support explicit relations about the social sphere. Kokar
include technical details. For in-depth accounts, the reader is et al. [2009] acknowledge that their framework is a starting
referred to Kokar et al. [2009] and Fiske [1992]. point, thus it is not complete and is open for extensions. We
do so by using concepts from sociology presented below.
3.1 Situation Theory Ontology
Kokar et al. [2009] present a formalization of situations 3.2 Elementary Forms of Sociality
based on the situation theory developed by Barwise and Perry Fiske [1992] in his attempt towards a Unified Theory of So-
[1981] and extended by Devlin [1995]. This formalization is cial Relations suggests that most kinds of social interaction
compatible with the interpretation of situation awareness in are generated by using four elementary relational models.
terms of human awareness provided by Endsley [2000]. They More complex social forms can then be expressed by combin-
formalize these concepts by using the Web Ontology Lan- ing these models. The relational models are not dependent on
guage [W3C, 2004]. The authors suggest that an agent can specific domains or situations, rather they can be used for any
be considered aware of the situation if it has “...data pertinent type of social relations. This feature makes the approach use-
to the objects of interest, some background knowledge that ful for our cause, since we aim at having a generic framework
allows one to interpret the collected object data and finally which can model any type of (social) situations.
a capability for drawing inferences”. Furthermore, they ar- Communal Sharing (CS) relationships represent a
gue that the agent should not only reason about relations, but bounded group of people who are equivalent and undiffer-
also recognize situations and how they impact on one’s goals. entiated. An example of this relation is an open buffet dinner:
According to the assumption by Barwise and Perry [1981], the guests can all take as much food as they want, so they are
situations are simply limited parts of the world perceived by in some sense indistinguishable. However, these equivalence
people. Devlin [1995] emphasizes the importance of under- classes are not fixed: people in the same equivalence class
standing that the information that the agent has is simply a are seen as equal only for that specific purpose. Formally, CS
part of the overall theoretically available information. is an equivalence relationship in which reflexivity, symmetry,
The main elements of situation theory are objects and and transitivity hold.
types. Some of the basic types are IND representing the type Authority Ranking (AR) relationships represent a hierar-
of individuals, RELn representing n-place relations, SIT rep- chical ordering of people for a certain social dimension. Peo-
resenting the types of situations etc. The main type of situ- ple in higher ranks have more power, privileges, and typically
ation is an utterance situation, representing an expression in also more responsibilities. An example of this relation can be
natural language which is linked to a real situation. These found in a military setting. Formally, AR is a linear ordering,
utterances can be represented as information coming from a i.e., a reflexive, transitive and anti-symmetric relationship.
user. The relevant part of the world mentioned in the utter- Equality Matching (EM) relationships represent a model
ance is called focal situation. Utterances might also refer to of even balance, for example turn taking, tit-for-tat retalia-
another situation, called resource situations, which is used as tion, or compensation by equal placement. People are mainly
background information during the reasoning process. concerned whether there is balance in the relationships. An
The basic information about a situation is expressed by example can be parents taking turns to baby-sit their chil-
using infons, written as: << R, a1 , ..., an , 0/1 >> where dren. Formally, the EM relationship has the properties of a
R is an arbitrary n-place relation, a1 , ..., an are objects ap- linear ordering and entails the idea of an additive identity.
propriate for R, and 0/1 is the polarity of the infon, show- Furthermore, the relation obeys the associative and commu-
ing whether the relation holds for those objects or not. For tative laws. Finally, EM is order preserving.
instance, the infon << workT ogether, Alice, Bob, 1 >> Market Pricing (MP) relationships represent a model of
would express that the relation workT ogether(Alice, Bob) proportionality in social relationships. All the features of the
holds. Situations and infons are related by the support rela- relation are considered under the point of view of a single
tionship (), which relates the situation with infons that hold utility metric that allows for comparison. Typical examples
31
Tenth International Workshop Modelling and Reasoning in Context (MRC) – 13.07.2018 – Stockholm, Sweden
of MP relations are those which include a rational calculation If s << member, X, G1, 1 >>
of expected utility, where the usual metric of comparison is and s << communalSharing, G1, 1 >>
money. Formally, MP relationships have the properties of EM and s << member, X, G2, 1 >>
and some others such as the concept of multiplication as well and s << authorityRanking, G2, 0 >>
as the fact that two entities of the same relational structure Then s << equal, choice(X), pref (X), 1 >>
can be compared.
which shows that if a person is not a member of some AR
Dignum [2004] proposes three types of relationships which
group (i.e., there is no choice conflict), then the choice of the
govern transactions in organizations: a hierarchical model,
person can be the same as her preference.
a market model [Williamson, 1975], and a network model
[Powell, 2003]. These are similar to the AR, MP, and a com- [Behind the scenes of s1] In this situation the relevant
bination of CS and EM relations, respectively. individuals are Alice, Bob and Charlie, and the rele-
vant relation is CS since we are talking about a group
4 Suggested Approach of friends going on holidays, so we can say that
In order to represent the example of Section 2, we split the s1 << member, Alice, f riendsGroup, 1 >>
scenario into different situation snippets:
s1 << member, Bob, f riendsGroup, 1 >>
s1 Alice, Bob and Charlie have to pick a time and place for
their holiday trip; s1 << member, Charlie, f riendsGroup, 1 >>
s2 Alice notices she has a meeting with her boss on the de- s1 << communalSharing, f riendsGroup, 1 >>
parture day; Furthermore, at this point we do not have information
s3 Alice asks Bob for a beer; about conflicting choice due to Alice being in an AR
group. By using the inference rule, the system under-
s4 Alice buys a chocolate in the shop;
stands that in order for a decision to be made, first of all
s5 Alice gives the chocolate to Bob. relevant individuals should be asked, and once there is
In order to model these situations, we assume that the agent no disagreement, a place (and time) can be picked.
has knolwedge about Alice’s social relations. How to build
these knowledge structures with the users will be explored In AR, priority goes to people holding higher positions. In
in future work. As aforementioned, we will extend the STO this spirit, we can define the rules:
by adding social relations and inference rules. At the same If s << member, X, G, 1 >>
time, we will explain how our situation snippets can be mod- and s << member, Y, G, 1 >>
elled using these rules and relations. Since we are in an ontol- and s << higherAuthority, X, Y, G, 1 >>
ogy setting, the presented relations are instances of the RELn Then s << outRanked, Y, G, 1 >>
type, while all the individuals are instances of the IND type.
We do not express the overall ontology due to space restric- If s << authorityRanking, G, 1 >>
tions, however for each situation snippet we present the rel- and s << member, X, G, 1 >>
evant individuals and relations, as well as the infons that are and s << outRanked, X, G, 0 >>
supported by the situation. Finally, we express what can be and s << member, Y, G, 1 >>
derived from that information in combination with the infer- and s << outRanked, Y, G, 1 >>
ence rules. Then s << equal, select(G), choice(X), 1 >>
In CS relations, decisions are made by unanimous agree- and s << equal, choice(Y ), select(G), 1 >>
ment by the members. This bit of knowledge is represented
by the following inference rules: The last rule suggests that if a person is outranked in an
AR relation, then its choice is the same as the choice of the
If s << member, X, G, 1 >> top member of the group.
and s << member, Y, G, 1 >>
and s << equal, X, Y, 0 >>
and s << equal, choice(X), choice(Y ), 0 >> [Behind the scenes of s2] In this situation the relevant
Then s << disagree, G, 1 >> individuals are Alice and her boss. Bob and Charlie are
not relevant for the situation at this point, since we are
where G represent a group of people that X is a part of, taking the point of view of Alice. The relevant situation
is AR, and the situation supports the following infons:
If s << communalSharing, G, 1 >> s2 << member, Alice, workGr, 1 >>
and s << disagree, G, 0 >>
and s << member, X, G, 1 >> s2 << member, boss, workGr, 1 >>
Then s << equal, selection(G), choice(X), 1 >> s2 << authorityRanking, workGr, 1 >>
this means that if there is no disagreement in the group, the s2 << higherAuthority, boss, Alice, workGr, 1 >>
choice of any member can be selected. s2 << outRanked, boss, workGr, 0 >>
32
Tenth International Workshop Modelling and Reasoning in Context (MRC) – 13.07.2018 – Stockholm, Sweden
so the systems concludes that Alice is outranked and In future work we will model this by formulating a meta or
therefore has to attend the meeting with her boss. second-order property of predicates: persistent(P ) means
that P remains true unless explicitly changed by the rules on
At this point, s1 will also change, since now Alice’s choice the object level.
is different (and is not the same as her preference), therefore In an MP relation, money is the metric through which an
there is disagreement in the group. This would give rise to exchange balance is kept, so we introduce the following rule:
a new situation where the relevant individuals are Alice, Bob
and Charlie, and s2 would be a resource situation, but we will If s << marketP ricing, X, Y, 1 >>
not model it for space reasons. and s << serves, Y, X, V, 1 >>
In EM, the most important aspect is to keep some kind of Then s << shouldP ay, X, Y, val(V ), 1 >>
balance in the relation. We can define the following rules:
[Behind the scenes of s4] In this situation the relevant
If s << equalityM atching, X, Y, 1 >>
individuals are Alice and the shopkeeper, and the rele-
and s << owesF avor, X, Y, V, 1 >>
vant relation is MP. The following infons are supported:
and s << serves, X, Y, W, 1 >>
Then s + 1 << owesF avor, X, Y, val(V ) − val(W ), 1 >> s4 << marketP ricing, Alice, shopkeeper, 1 >>
Here s + 1 refers to the situation that follows from s by that s4 << serves, shopkeeper, Alice, chocolate, 1 >>
actions taken in s, V and W represent the owed items, while so the system can infer that Alice should pay for the
the function val represents the value of the items. chocolate.
If s << owesF avor, X, Y, val(V ), 1 >> At this stage MP seems like a sub-case of EM, however the
Then s << owesF avor, Y, X, −val(V ), 1 >> difference stands in the fact that EM usually implies a closer
relation between the people, and the concept of paying back
If s << owesF avor, X, Y, 0, 1 >> a favor is more subjective, while in MP the values are usually
Then s << balanced, X, Y, 1 >> clear to all parties.
If s << balanced, X, Y, 1 >> 5 Conclusions and Future Work
Then s << balanced, Y, X, 1 >> In order to support people in their daily activities, artificial
agents should not only be able to reason about the user’s inter-
[Behind the scenes of s3] In this situation the relevant nal processes, but at the same time they should understand the
individuals are Alice and Bob, and the relevant relation user’s surrounding environment. In this paper we focus on the
is EM a . The following infons are supported: social environment, and we present a framework that can be
used to model situations which include social elements. Fur-
s3 << equalityM atching, Alice, Bob, 1 >> thermore, we illustrate how our framework functions through
s3 << serves, Bob, Alice, beer, 1 >> an example. As our starting point we use the Situation The-
ory Ontology developed by Kokar et al. [2009], and we ex-
so from now on Alice should keep in mind that she owes tend it in order to enable it to represent social relations. We
a favor to Bob until she has repaid him. implement four basic social relations: communal sharing, au-
a
This situation can also be modelled as CS, if that is Alice’s thority ranking, equality matching and market pricing [Fiske,
subjective view on her relation with Bob. In that case no rule 1992]. The resulting framework allows us to represent the so-
would trigger. We assume EM for illustration purposes. cial aspects of situations, and is able to reason based on these
situations. The inference rules are based on properties of the
[Behind the scenes of s5] In this situation the relevant social relations. A key feature of the framework is that it rep-
individuals are Alice and Bob, and s3 is a resource situ- resents situations from the point of view of the user that we
ation. Again, EM is the relevant relation, in which: want to support.
In future work, we will add more generic inference rules
s5 << equalityM atching, Alice, Bob, 1 >> based on societal norms in order to enable the system to rea-
son about arbitrary situations. Furthermore, the approach can
s5 << serves, Alice, Bob, chocolate, 1 >> be combined with activity recognition technology in order to
in s3 we inferred that Alice owes a favor to Bob. In Al- enable encoding the infons automatically. This way the agent
ice’s point of view a beer and a chocolate have a similar can use as an input explicit knowledge from the user as well
value, so the system can conclude that she does not owe as knowledge from the recognition module, and then in turn
something to Bob anymore: the balance has been reset. reason and act based on the preferences of the user. More-
over, we will include the temporal aspect of situations which
For this to work in the formalism, we need to be able to re- is currently missing. Another goal is to unify our framework
fer that s5 is later than s3 and that if Alice did not serve Bob with a framework that accounts for the user’s behaviour, this
with something in the mean time, she still owes him the same way both the internal and external aspects are represented.
as she did in the state following from s3. This inertia/per- Finally, we want to make it possible for users to build the
sistence of (sub-)situations is known as the frame problem. representations of situations interactively with the system.
33
Tenth International Workshop Modelling and Reasoning in Context (MRC) – 13.07.2018 – Stockholm, Sweden
Acknowledgements M. Birna Van Riemsdijk, Catholijn M. Jonker, and Victor
This work is part of the research programme CoreSAEP, with Lesser. Creating Socially Adaptive Electronic Partners: In-
project number 639.022.416, which is financed by the Nether- teraction, Reasoning and Ethical Challenges. In Proceed-
lands Organisation for Scientific Research (NWO). The au- ings of the 2015 International Conference on Autonomous
thors would like to thank the anonymous reviewers for their Agents and Multiagent Systems, pages 1201–1206. Interna-
valuable comments and suggestions. tional Foundation for Autonomous Agents and Multiagent
Systems, 2015.
References W3C. Web Ontology Language Reference OWL, 2004.
Varol Akman and Mehmet Surav. Steps toward formalizing Oliver E. Williamson. Markets and Hierarchies: Analysis and
context. AI magazine, 17(3):55, 1996. Antitrust Implications. New York: Free Press, 1975.
Jon Barwise and John Perry. Situations and Attitudes. The
Journal of Philosophy, 78(11):668–691, 1981.
Jon Barwise. Situations and Small Worlds. Handbook of
Semantics, 1987.
Keith Devlin. Logic and Information. Cambridge University
Press, 1995.
Virginia Dignum, Catholijn M. Jonker, Rui Prada, and Frank
Dignum. Situational Deliberation; Getting to Social Intelli-
gence. Computational Social Science and Social Computer
Science: Two Sides of the Same Coin, 2014.
Virginia Dignum. A Model for Organizational Interaction:
Based on Agents, Founded in Logic. SIKS, 2004.
Mica R. Endsley. Theoretical Underpinnings of Situation
Awareness: A Critical Review. Situation Awareness Anal-
ysis and Measurement, 1:24, 2000.
Alan P. Fiske. The Four Elementary Forms of Sociality:
Framework for a Unified Theory of Social Relations. Psy-
chological Review, 99(4):689, 1992.
Gal A Kaminka. Curing robot autism: A challenge. In
Proceedings of the 2013 international conference on Au-
tonomous agents and multi-agent systems, pages 801–
804. International Foundation for Autonomous Agents and
Multiagent Systems, 2013.
Mieczyslaw M. Kokar, Christopher J. Matheus, and Kenneth
Baclawski. Ontology-based Situation Awareness. Infor-
mation Fusion, 10(1):83–98, 2009.
Karen L Myers and Neil Yorke-Smith. Proactivity in an in-
tentionally helpful personal assistive agent. In AAAI Spring
Symposium: Intentions in Intelligent Systems, pages 34–
37, 2007.
Robert Ezra Park and Ernest Watson Burgess. Introduction
to the Science of Sociology. University of Chicago Press
Chicago, 1921.
Walter Powell. Neither Market nor Hierarchy. The Sociol-
ogy of Organizations: Classic, Contemporary, and Critical
Readings, 315:104–117, 2003.
Herbert A. Simon. Theories of Bounded Rationality. Deci-
sion and Organization, 1(1):161–176, 1972.
Myrthe Tielman, Willem-Paul Brinkman, and Mark A Neer-
incx. Design guidelines for a virtual coach for post-
traumatic stress disorder patients. In International Con-
ference on Intelligent Virtual Agents, pages 434–437.
Springer, 2014.
34