=Paper=
{{Paper
|id=Vol-1660/womocoe-paper4
|storemode=property
|title=Interaction Specifications as Contexts for Ontologies
|pdfUrl=https://ceur-ws.org/Vol-1660/womocoe-paper4.pdf
|volume=Vol-1660
|authors=Paula Chocron,Marco Schorlemmer
|dblpUrl=https://dblp.org/rec/conf/fois/ChocronS16
}}
==Interaction Specifications as Contexts for Ontologies==
https://ceur-ws.org/Vol-1660/womocoe-paper4.pdf
Interaction Specifications
as Contexts for Ontologies
Position Paper
Paula CHOCRON a,b,1 and Marco SCHORLEMMER a
a
Artificial Intelligence Research Institute, IIIA-CSIC
Bellaterra (Barcelona), Catalonia, Spain
b
Universitat Autònoma de Barcelona
Bellaterra (Barcelona), Catalonia, Spain
Abstract
Common formalisms to represent knowledge, such as Description Log-
ics, are insufficient to represent the dynamic aspects of language when
it is used in communication. However, this information can be help-
ful to describe the semantics of terms in interactions, and in particular
to obtain useful alignments that can be used as translations between
heterogeneous interlocutors. We propose to consider the specification of
interactions as contexts for ontologies. To this aim, the ontology lan-
guage needs to be extended with a temporal layer that allows to express
properties about the possible message exchanges between agents in an
interaction. We present a simple preliminary extension for taxonomies,
and discuss directions to investigate.
1. Introduction
The problem of specifying interactions for artificial agents has been extensively
studied in the past decades. These efforts resulted in different formalisms to de-
scribe interactions, such as agent communication languages [14], or ontologies spe-
cially designed to this aim [8]. The semantics of these formalisms has been subject
of discussion for several years, mainly because it is not clear how to combine infor-
mation about the intentionality of communicating agents with the representation
of knowledge about the domain of conversation.
One approach to overcome these difficulties is the one presented in [13], where
the authors propose a layered semantics of communication languages as a way
to represent different aspects of meaning in interaction. A first layer, called the
content level, is internal to each agent, and represents the semantics of the con-
versation domain. This layer describes the meaning of the contents of messages,
independently from the interaction in which they are uttered. Description Logics
are a natural choice to specify this layer, since they provide a way to express prop-
1 Corresponding Author. E-mail: pchocron@iiia.csic.es
erties about concepts and their relations. A second layer, the action level, corre-
sponds to the specification of a particular interaction. In this layer, the meaning
of a message is determined by the effect of uttering it in a conversation, or more
concretely, by its possible answers. This layer is external to the agents, and can
be modelled by means of interaction protocols.2 The importance of interaction
contexts can be seen clearly in human conversations; consider for example a dia-
logue between a waiter and a customer. If the waiter says “what would you like
to drink?”, an adequate answer would be “wine”, while “potatoes” would likely
cause confusion.
In this position paper we propose to consider interaction specifications as
contexts for ontologies, extending the ontology representation language with a
temporal layer that allows to describe properties of a dialogue. This leads to
a formalism that is useful to reason about language in use, by modelling prag-
matic aspects of meaning that depend directly on the particular interaction tak-
ing place, and the way in which they relate to the underlying ontology. A lan-
guage relating ontological and pragmatic information could be useful to perform
ontology matching for the particular application of agent communication, since
the information on the interaction context can be very valuable when building an
alignment to be used by agents that interact to perform a particular task. In [3],
Atencia and Schorlemmer show how a shared interaction context can already be
sufficient for artificial agents to understand each other with a certain degree of
success, even if they do not share the underlying vocabulary; in [6] pre-existing
ontology alignments are incorporated to this idea, showing how the interaction
context can be used to fix alignments that can be incorrect and incomplete. How-
ever, in this last approach the information from the context is only considered
separately; we propose to develop a language that can be used to reason in the
combination of both levels. In Section 3 we discuss in particular its application
in ontology matching.
Unlike the context as a box approach to contextualizing ontologies [10], we do
not consider contexts to be a separate theory, but rather a temporal layer built
over the elements in the ontology. The original ideas about temporal extensions
of description logics were applied naturally to planning [1], but since this does
not involve communication with other agents, it does not need branching time
and uses LTL, a logic whose relation with description logics is well known [2].
In [15] the authors study two-dimensional logics of context, which also extends a
DL-based language to describe dynamic aspects of knowledge. In our approach,
the temporal layer is only used to describe how elements in the ontology are
structured in a particular dialogue. To model interactions between agents that
make decisions autonomously, we need to incorporate branching-time operators,
in order to represent the possible messages agents can decide to utter. The relation
between CTL and description logics has only been studied recently and results
about decidability are not very hopeful [11]; we later discuss why this does not
necessarily apply in our case.
2 Pitt and Mandami include a third layer, which corresponds to the intention of communicating
of each agent. We will focus on the first two.
2. Models of Interaction as Context
We consider interactions between agents that communicate to perform a partic-
ular task together (for example, ordering drinks). The messages that agents send
to each other are predicates in a content language that may allow to describe
properties about the domain of the conversation. In this position paper, for sim-
plicity, we will only consider a simple taxonomy of concepts with an inclusion (v)
relation, but extensions to more complex formalisms (such as Description Logics)
should be considered.
Interaction contexts are specifications of possible message exchanges between
agents. The question of how it is best to model interaction protocols has been
extensively discussed in the multi-agent community, and many options with dif-
ferent expressive power were proposed [7]. To keep interaction models as general
as possible, we specify only possible interchanges of messages by means of Fi-
nite State Automata. Therefore a model of interaction (or context) is an FSA
that has its transitions labelled with messages in the content language along with
identification of the sender and receiver agents.
Example 1. The FSA in Figure 1 specifies an interaction for a waiter that com-
municates with a customer (c) who wants to order. State transitions should be
read as (sender, receiver) : message. Suppose also that the waiter represents in-
formation about the domain with a taxonomy that includes the relations Beer v
Drink, Water v Drink, Wine v Drink.
10
(c, w): Pint
(w, c): Size (c, w): Half pint
2 8 9
(c, w): Beer
(w, c): Drink (c, w): Wine (w, c): Color (c, w): White
start 0 1 3 5 7
(c, w): Water (c, w): Red
4 6
Figure 1. Interaction model for ordering drinks
A language to talk about interaction contexts should allow to express prop-
erties about the temporal structure of the automaton. In our example, one could
want to say that if the waiter asks the customer what she wants to drink, the
customer can answer that she wants a beer. More generally, we are interested in
expressing properties like the following ones:
- something is a possible answer in a particular state
- something can or must be said eventually from a particular state
To this aim, it is necessary to define a context language that extends the
content language describing these aspects of a particular interaction. A natural
way to express properties about Finite State Automata is to use the µ-calculus
formalism, but we do not need all its expressive power. For properties like the ones
above, it is enough to consider only the CTL operators that express necessity (A)
and possibility (E), together with the ones for next state ( d) and eventually (3).
We also need to include in the signature a set of agent names A and a predicate
utter, that relates two agents with a term in the ontology. If C is a concept in the
content language and a, a0 ∈ A, well-formed terms in the context language are
E d utter(a, a0 , C), E 3 utter(a, a0 , C), A 3 utter(a, a0 , C), A d utter(a, a0 , C),
utter(a, a0 , C). We consider a very simple semantics for the utter predicate:
utter(a, a0 , C) is true if agent a has sent message C to a0 . More complex under-
lying languages may need other predicates. For example, for first-order logics it
would be desirable to identify a query from an answer, for which two predicates
(commonly ask and tell) are necessary.
With the context language defined in this way, we can now express the fact
that Beer is a proper answer to Drink with the following predicate:
utter(w, c, Drink) =⇒ E d utter(c, w, Beer)
The language allows for interesting combinations with description logics that
should be analysed. For example, by introducing a new concept name D playing
the role of a fresh variable, we can express properties such as if the waiter asks
to the customer what she wants to drink, the customer will answer specifying a
drink in the following way:
utter(w, c, Drink) =⇒ A d utter(c, w, D) ∧ D v Drink
Note that, unlike previous work, we do not apply temporal operators directly
to elements of the underlying domain language, but rather to a predicate that
states who utters a message to whom. Therefore the results presented in [11] about
the complexity of CTL and DL should be reconsidered for this case in particular.
While our interaction models only describe who sends messages and when,
other kinds of information, such as preconditions and effects of utterances, are
also relevant in a conversation. However, this information can be encoded in the
underlying logics, if one uses, for example, a combination of description logics and
action formalisms [4] or knowledge and action bases [12]. The consequences of
considering this aspects as part of the context or as part of the content knowledge
should be investigated.
3. Alignment Between Contextualized Ontologies
One prominent application of using interaction specifications as context is to
build alignments between vocabularies of different agents that are adequate for
particular tasks. As argued in [5], maintaining standard ontologies in open and
distributed environments is a difficult task, particularly when they represent dy-
namic domains. Although there exist multiple ontology alignment tools that find
translation between vocabularies, they are not particularly designed to be used
10
(c, w): Media
(w, c): Quantità (c, w): Piccola
2 8 9
(c, w): Birra
(w, c): Bibita (c, w): Vino (w, c): Tipo (c, w): Bianco
start 0 1 3 5 7
(c, w): Aqua
(c, w): Rosso
4 6
Figure 2. Italian interaction model for ordering drinks
for agent communication, and therefore do not take into account the second layer
of meaning [9]. On the other side, as mentioned in the introduction, there ex-
ist approaches to align vocabularies using only the interaction models, without
taking into account any underlying knowledge [3].
To illustrate the problem of aligning vocabularies in the context of a particular
interaction, consider again the ordering drinks example, but we now suppose
that, while the waiter speaks English, the customer is Italian and follows the
interaction model in Figure 2. We consider agents that speak different languages
but share a basic knowledge of how to perform the task, which is represented
as having interaction protocols with similar structure (or equal in this case, for
simplicity). This gives place to an alignment between their vocabularies that
allows agents to interact successfully, for example, the term Piccola should be
in this case interpreted as Half Pint. While this equivalence is reasonable in the
context of ordering drinks, general ontology matchers would likely never find it
without additional information, such as the fact that Piccola is a proper answer
to Quantità and Half pint to Size. Our approach to contextualizing ontologies
by describing properties of a particular interaction can be used, on one side, to
specify this kind of pragmatic properties that can be clues for the alignment, and
on the other, to describe the contexts for which an alignment is valid.
Acknowledgements This research has been funded by the European Communi-
tys Seventh Framework Programme (FP7/2007-2013) under grant agreement no.
567652 /ESSENCE: Evolution of Shared Semantics in Computational Environ-
ments/.
References
[1] Alessandro Artale and Enrico Franconi. A temporal description logic for reasoning about
actions and plans. J. Artif. Int. Res., 9(1):463–506, August 1998.
[2] Alessandro Artale and Enrico Franconi. A survey of temporal extensions of description
logics. Annals of Mathematics and Artificial Intelligence, 30(1-4):171–210, 2000.
[3] Manuel Atencia and W. Marco Schorlemmer. An interaction-based approach to semantic
alignment. J. Web Sem., 12:131–147, 2012.
[4] Franz Baader, Carsten Lutz, Ulrike Sattler, and Frank Wolter. Integrating description
logics and action formalisms: First results. In Proceedings of the Twentieth National
Conference on Artificial Intelligence (AAAI-05), pages 572–577, 2005.
[5] Alan Bundy and Fiona McNeill. Representation as a fluent: An ai challenge for the next
half century. IEEE Intelligent Systems, 21(3):85–87, 2006.
[6] Paula Chocron and Marco Schorlemmer. Attuning ontology alignments to semantically
heterogeneous multi-agent interactions. In Proceedings of the 22nd European Conference
on Artificial Intellignce (ECAI), 2016 (to appear).
[7] Amit Chopra and Munindar P. Singh. Agent Communication. In Multiagent Systems: A
Modern Approach to Distributed Artificial Intelligence, pages 101–141. 2013.
[8] Stephen Cranefield, Martin Purvis, Mariusz Nowostawski, and Peter Hwang. Ontologies
for Agents: Theory and Experiences, chapter Ontologies for Interaction Protocols, pages
1–17. Birkhäuser Basel, Basel, 2005.
[9] Jérôme Euzenat and Pavel Shvaiko. Ontology Matching. Springer-Verlag New York, Inc.,
Secaucus, NJ, USA, 2007.
[10] Chiara Ghidini and Fausto Giunchiglia. Local models semantics, or contextual reason-
ing=locality+compatibility. Artificial Intelligence, 127(2):221 – 259, 2001.
[11] Vı́ctor Gutiérrez-Basulto, Jean Christoph Jung, and Thomas Schneider. Lightweight tem-
poral description logics with rigid roles and restricted tboxes. In Proceedings of the
Twenty-Fourth International Joint Conference on Artificial Intelligence, IJCAI 2015,
Buenos Aires, Argentina, July 25-31, 2015, pages 3015–3021, 2015.
[12] Babak Bagheri Hariri, Diego Calvanese, Marco Montali, Giuseppe De Giacomo, Riccardo
De Masellis, and Paolo Felli. Description logic knowledge and action bases. J. Artif. Intell.
Res., 46:651–686, 2013.
[13] Jeremy Pitt and Abe Mamdani. A protocol-based semantics for an agent communication
language. In Proceedings of the 16th International Joint Conference on Artifical Intel-
ligence - Volume 1, IJCAI’99, pages 486–491, San Francisco, CA, USA, 1999. Morgan
Kaufmann Publishers Inc.
[14] Stefan Poslad. Specifying protocols for multi-agent systems interaction. ACM Trans.
Auton. Adapt. Syst., 2(4), November 2007.
[15] Frank Wolter and Michael Zakharyaschev. Multi-dimensional description logics. In Pro-
ceedings of the 16th International Joint Conference on Artifical Intelligence - Volume 1,
IJCAI’99, pages 104–109, San Francisco, CA, USA, 1999. Morgan Kaufmann Publishers
Inc.