=Paper=
{{Paper
|id=Vol-2518/paper-WINKS1
|storemode=property
|title=Argumentation on Meaning: a Semiotic Model for Contrast Set Alignment
|pdfUrl=https://ceur-ws.org/Vol-2518/paper-WINKS1.pdf
|volume=Vol-2518
|authors=Kemo Adrian,Enric Plaza
|dblpUrl=https://dblp.org/rec/conf/jowo/AdrianP19
}}
==Argumentation on Meaning: a Semiotic Model for Contrast Set Alignment==
https://ceur-ws.org/Vol-2518/paper-WINKS1.pdf
Argumentation on Meaning: a Semiotic
Model for Contrast Set Alignment
Kemo ADRIAN a,b and Enric PLAZA a
a IIIA-CSIC, Campus UAB, Carrer de Can Planes, 08193 Bellaterra, Catalonia, Spain
b Universitat Autonoma de Barcelona, 08193 Bellaterra, Catalonia, Spain
Abstract. Argumentation theory has been investigated as a possible way to link
ontologies and machine learning, in multi-agent systems facing situations of het-
erogeneous knowledge. We are investigating scenarios where this argumentation
takes place over one contrast set, a segment of two agents’ ontology. Addressing a
whole contrast set is challenging, as argumentation theory is normally used to align
one pair of concepts from a contrast set. Our approach to use argumentation theory
to align a whole contrast set goes in three parts. First, we clarify and define what
is an acceptable alignment between two contrast sets with the notion of agreement.
Then, we formalize as disagreements the pairing relations between concepts that
prevent this agreement. Finally, we propose a model that identifies and resolves
disagreements. This article focuses on the presentation of our model together with
a preliminary experimental evaluation.
Keywords. Ontologies, Machine Learning, Inductive Learning, Argumentation
Theory, Multi-Agent System.
1. Introduction
In machine learning, agents can learn to classify elements – called object or examples –
from their environment. This learning gives to the agents a structured knowledge about a
domain, a knowledge which is individual to each learning and therefore to each agent. A
classification can be seen as a contrast set of an ontology, a particular subset of the ontol-
ogy regrouping a set of classes or concepts partitioning a specific domain. For instance,
the colors partition the visible spectrum domain of any ontology including it. Therefore,
aligning the classes of a same domain from different ontologies becomes similar to align-
ing two classifiers. The classifiers, however, have the support of their learning data-sets
to complement their ontological knowledge. An approach that takes advantage of these
two aspects in the task of classifier alignment is the formal argumentation theory, where
both the examples and the generalizations learned over these examples can be used as
arguments in order to create new generalizations which allow consistent classifications
over larger contexts.
Ontology alignment and machine learning both address the question of meaning.
There approaches to that question, however, are different. In previous work [2], we al-
ready discussed that machine learning tends to have an externally grounded approach
of meaning [5][11], while ontology alignment can either consider meaning as a concep-
tual web [6] or externally grounded. Argumentation has often been used in a conceptual
web approach of ontology alignment [12][7], but recent work on interaction based on-
Copyright c 2019 for this paper by its authors. Use permitted under Creative Commons License Attribution
4.0 International (CC BY 4.0).
Figure 1. Association Paths as Semiotic Triangles.
tology alignment also uses grounded concepts [4]. We aim to provide an argumentation
framework in the latter paradigm, which focuses on limiting the information exchanged
between the agents.
A classifier alignment can be conceptualized as a naming game between two agents,
where they both need to name the examples from a set U called their context, achieving
the highest number of examples named similarly. In order to do so, the agents need to
be able to associate each example from the context with a sign from a lexicon. We dis-
tinguish two methods for an agent to map a context U to a lexicon L: the right-path and
the left-path associations. The right-path associations of a classifier K, noted U Kr 7→L, are
received from an external source by the agent, for instance from the experimenter as a
training data-set. They consist of an array of example-sign pairs. The left-path associ-
ations of a classifier K, noted U Kl 7→L, however, use the knowledge of the classifier to
associate a sign with an example. The example is fed to the classifier as an input, and
the same classifier outputs its sign. Two right-path associations can map a same exam-
ple to different signs, creating inconsistencies in the knowledge of the agents. Moreover,
only examples from the context known by an agent can be named through right-path as-
sociations. A same left-path association, however, can name new examples through the
process of classification. Moreover, the result of a classification being reproducible, two
agents classifying with the same generalizations are guaranteed to classify in a consistent
way and score high in their naming game.
Argumentation frameworks are used in artificial intelligence to manage contentious
information among multi-agent systems and draw conclusions from it. Among them, the
framework AMAIL [8] has been shown to be a promising way to align two classifiers
which are using a feature structures formalism to represent their knowledge. In order
to align two classifiers, the AMAIL framework uses argumentation over each pair of
concepts that belong to different agents but are supposed to be equivalent in order to
find a new common set of generalizations for the two agents to have similar left-path
associations. Unfortunately, this means that the AMAIL algorithm needs to know which
concepts are expected to be equivalents in order to align them. Moreover, the AMAIL
algorithm focuses on an approach based on the prevalence of right-path associations,
which make the framework unable to work on scenarios with inconsistent data, where
there are unique examples associated to different signs through the right-path.
Figure 1 shows that these two paths, which are linking signs, generalizations and
examples draw a triangle that evokes a semiotic triangle [10], which is why we are de-
scribing our model to align classifiers as a semiotic model. The elements of our model
are named from their semiotic equivalents, and they are used to formalize the notion
of classifier alignment into the partitioning, pairing and adjustment of fairly-equivalent
concepts from different contrast sets. The notion of alignment becomes an agreement
over the meaning, as the agents use argumentation on each of these pairs of concepts in
order to align them. By introducing the notion of pairing relations, we allow the agents to
use an argumentation framework inspired from AMAIL that can be used without a pre-
existing mapping of the expected equivalence relations between the concepts of the dif-
ferent agents. Moreover, we adopt a constructivist approach on argumentation, where the
left-path associations prevails over the right-path associations, which allows the model
to work on scenarios with inconsistent data. The full formalism of our approach has been
formalized in a synthetic article [2]. Section 2 presents a short summary of this article.
2. Approach
In our approach, objects – called examples – are represented as feature terms [3]. Feature
terms, also called feature structures or ψ-terms, are a generalization of first-order terms
that have been introduced in theoretical computer science in order to formalize object-
oriented capabilities of declarative languages. Feature terms correspond to a different
subset of first-order logic than description logics, but have a similar expressive power.
Generalizations of these examples can also be represented as feature terms. The basic
operation between feature terms is subsumption: we will use ψ1 v ψ2 to express that a
term ψ1 subsumes another term ψ2 – that is to say ψ1 is more general (or equal) than
ψ2 . Our approach is broadly inspired by semiotics, and and we will now introduce some
basic notions of that approach.
2.1. Semiotic Elements
The semiotic elements are the building blocks of our knowledge representation. They
help us to formalize the notions of example-sign associations, consistency and align-
ment. The first three semiotic elements are the extensional definition E, the intensional
definition I and the sign s. An extensional definition is a set of examples, an intensional
definition is a set of generalizations and a sign is just a symbol (a string of characters in
our case). These three semiotic elements can be combined into a concept C, which is the
fourth and last type of semiotic elements. A concept is composed of an extensional, an
intensional definition, and a sign, such that the intensional definition subsumes the exam-
ples of the extensional definition. The semiotic elements of the concept C are s(C), I(C)
and E(C) and we note C = hs(C), I(C), E(C)i. In a concept, the association between s(C)
and E(C) represents the right-path associations, while the associations between s(C) and
an example subsumed by a generalization g ∈ I(C) are the left-path associations.
2.2. Contrast Set
A single concept only partitions a subset of a context U through its associations. In
order to partition a whole context, the agents need a set of concepts. A contrast set
K = hU, S = {C1 , . . . ,Cn }i is a relation between a context U and the set of concepts
{C1 , . . . ,Cn } such that these concepts partition U, meaning that each example in U is
subsumed by one and only one concept from S; this means any intensional definition I(C)
does not subsume any example in another E(C0 ) whenever C,C0 ∈ S. A variation of the
contrast set is the hypothesis. A hypothesis H is a contrast set without the constraint on
the relation between the context and the set of concepts. A hypothesis is usually used by
one agent A1 to project the left-path associations of another agent A2 on its own context
U1 . Therefore, the set of concepts of the other agent may or may not partition U1 , and
therefore may or may not be a contrast set.
2.3. Pairing Relations
A pairing relation qualifies the relation between two concepts in a given context. A pair-
ing relation is independent of the extensional definitions of the concepts, as the left-path
associations are used to compute the pairing relations between concepts. The process of
computing a pairing relation is presented with more details in our previous paper [2].
The pairing relation between two concepts C1 and C2 in a context U is computed by find-
ing these two concepts’ adjunct sets Ad j(C1 ,U) and Ad j(C2 ,U) such that adjunct set is
Ad j(Ck ,U) = {e ∈ U|I(Ck ) v e}. From these adjunct sets, three pairing partial sets are
obtained: U1, 2 , U 1,2 and U1,2 . The two firsts sets are the two set differences of Ad j(C1 ,U)
and Ad j(C2 ,U), while the third set is the intersection of Ad j(C1 ,U) and Ad j(C2 ,U). The
three pairing partial sets are defining a r-triplet, a triplet of Boolean values r(C1 ,C2 ,U)
that represents whether or not each pairing partial set is empty. Each of the 23 possible r-
triplets is mapped to a particular pairing relation: namely, overlap, inclusion, equivalence
and disjunction are the four main types of pairing relations.
The main advantage of this model of computation is the fact that we can obtain the
overall r-triplet r(C1 ,C2 ,UO ) of two concepts C1 and C2 in the overall context of our
agents, by applying a composition law to the agents’ two local r-triplets r(C1 ,C2 ,U1 ) and
r(C1 ,C2 ,U2 ). This composition law allows the agents to understand how their concepts
are semantically related at the general level without having to exchange their examples.
In the first version of our model, we were only using Boolean values in our r-triplets to
represent the emptiness of each pairing partial set. This did not allowed us to consider
cases where the intensional definition learned for a concept was not exactly subsuming
the concept’s extensional definition. In order to manage a certain degree of error when the
agents are using inductive learning, the r-triplets are now containing the cardinal number
of each pairing partial set. This modification has consequences on the composition law,
but inferring overall triplets from local triplets remains useful to limit the number of
examples exchanged by the agents.
2.4. Agreement on Meaning
The aim of an argumentation is to have the two agents involved in the argumentation
reach a state of mutual intelligibility. This state is reached when both agents are classi-
fying any given example of their overall context in two concepts that share a same sign.
When an agent is convinced that this state has been reached, we say that this agent is
in synchronic agreement. Besides the mutual intelligibility, the agents also want to still
be able to discriminate, after the argumentation, the examples that they originally were
classified in different concepts. If the new contrast set of an agent is, indeed, a refine-
ment of the old one, we say that this agent is in diachronic agreement. Before an argu-
mentation, an agent is always in diachronic agreement but not in synchronic agreement.
The argumentation on meaning helps the agents to reach a synchronic agreement without
breaking the diachronic agreement.
The synchronic agreement can be expressed as a set of conditions on the overall
pairing relations of the two agents. A violation of these conditions is called a synchronic
disagreement. We list seven types of synchronic disagreements, divided in four families.
The overlap disagreement and the hypo/hypernymy disagreement are semantic disagree-
ment, caused by concepts that have nor disjoint neither equivalent concepts in the other
contrast set. The synonymy and the homonymy disagreements are lexical disagreements,
respectively caused by equivalent concepts that are not sharing their sign and disjoint
concepts that are sharing the same sign. An indiscernible disagreement is caused by a
relation between two concepts that is left with too few proper examples on one concept.
A self disagreement is caused by two overlapping concepts from the same contrast set.
Finally, the untranslatable disagreement is caused by a concept not having an equivalent
concept in another contrast set.
3. Model
Our two agent model of argumentation will tackle the disagreements between the agents
sequentially, reducing the number of disagreements while keeping a diachronic agree-
ment for both agents. Before the argumentation, each agent has an initial contrast set
that is learned over examples of a domain (a ‘data set’), including a set of right-path
associations (supervised learning). The argumentation takes place turn by turn, with one
agent receiving a token at the beginning of the argumentation, taking a set of actions,
then passing the token to the other agent. An agent can take actions only when it has the
token, and the last action that it takes is always to pass the token to the other agent. The
token is exchanged until termination is detected. The actions taken by an agent while it
has the token are decided by the agent’s inputs and its current state. The two variables
that impact the behaviour of one agent at a given turn are the agent’s state (a qualitative
variable) and the messages that this agent has received from the other agent. Each agent
has the same set of possible states, making our argumentation model symmetric.
3.1. General functions of an agent
Our agents have 8 principal functions. Their first and basic function is their ability to
send elements of their knowledge (semiotic elements, r-triplets, pairing relations, etc.) to
the other agent through messages. The seven remaining functions are listed below.
Deleting Concepts. The agents can delete concepts from their contrast sets at anytime.
Since the agents are building hypotheses as an image of each others contrast sets, when
an agent A1 deletes a concept C from its contrast set it notifies the other agent A2 with a
message, asking A2 to delete C from its hypothesis.
Creating Concepts. The agents can create new concepts. The easiest way is to use a set
of right-path associations, and use them as a set of inputs for inductive learning. The in-
ductive learning generates a set of generalizations that become the intensional definition
of the new concept(s), while the lexicon of the set of associations becomes the sign(s)
of the new concepts and the examples become the extensional definition(s). Instead of
creating an intensional definition from an extensional definition, the agents can also cre-
ate an extensional definition from a set of generalizations. If an agent Ai with a context
Ui receives an intensional definition I associated to a sign s, it can create a new concept
C = hs, I, {e ∈ Ui |I v e}i. Finally, the agents can create concepts through argumentation.
The creation of a new concept C through argumentation triggers a secondary argumen-
tation within the argumentation on meaning, where the agents first define the properties
of the adjunct set Ad j(C,UO ) of the new concept in the overall context. Then, an agent
A1 uses induction to learn an intensional definition I of the new concept from the ad-
junct set Ad j(C,U1 ) over its local context U1 . I might be inaccurate, as the examples
from the other agent’s context are not taken into account during its creation, and in the
argumentation process the agents will modify I until both are satisfied. We do not detail
this argumentation process in this article, but it is fairly similar to the AMAIL algorithm.
Finally, the agents decide on a sign for the new concept.
Reassigning Overlaps. When a set of examples is left-path associated to two labels, the
agents can choose to reassign them to one of these two labels. Any method can be used
to reassign them, as long as this method assigns each example to the same label inde-
pendently of the agent using it. In our implementation, we are using the anti-unification
similarity measure [9] between each example that needs to be reassigned and the inten-
sional definitions of the two concepts that have the two involved labels for signs. Each
example is reassigned to the label linked to the intensional definition that has the highest
similarity with them.
Determining Overall Pairing Relations. In order to compute the overall pairing relation
between two concepts, the agents first compute their local adjunct sets. With them, they
compute the local pairing partial sets, then the local r-triplets. After exchanging their
local r-triplets, the agents can compute the overall pairing relation using composition
laws presented in our previous publication [2].
Finding disagreements. Once an agent knows about the overall pairing relations be-
tween its concepts and the other agent’s concepts, this agent can find the disagreements
between them as these disagreements are directly expressed as configurations of pairing
relations and signs. Using the overall pairing relations allow the agents to resolve their
disagreements over the whole set of examples that they collectively have access to, while
guarantying that the pairing relations used to qualify the disagreements are the same for
both agents —unlike local relations that are individual to each agent.
Looking for transitivity issues. When our model admits a degree of error, the properties
of the pairing relations change. The most problematic change is the lost of transitivity
for the relation of equivalence. Given three concepts A, B and C, if we had A ≡U B and
B ≡U C, then we had automatically A ≡U C with Boolean r-triplets. However, once we
introduce an error threshold, the relation between B and C is not necessarily an equiv-
alence anymore. To address this issue, the agents need to look for such configurations
with their overall pairing relations and delete either A or C. The concept that has the most
examples shared with B is the one deleted.
Reassigning lexicon items. The lexicon of an agent is the set of signs used in its current
contrast set. Once the agents have aligned their contrast sets in a configuration that allows
a one-to-one mapping of their concepts, they can reassign their lexicon items in order
to get rid of their lexical disagreements. Moreover, when concepts are created they have
new signs added into the lexicon. In order to reuse the old signs that became unassigned
following their concepts deletion, the agents can reassign them to the concepts that are
using new signs after reaching mutual intelligibility.
3.2. Resolving disagreements
Since disagreements are always related to pairs of concepts, removing one concept of
the two resolves the disagreement. However, the deletion can cause new disagreements
to appear. New disagreements caused by the resolution of explored disagreements are
addressed in the same way as any disagreement.
Semantic Disagreements. Semantic disagreements are resolved by deleting one of the
concepts that causes them and creating new concepts for the examples of the overall
context that become unassigned. A hypo/hypernymy is resolved by removing the hy-
pernym from its contrast set and replacing it by two co-hyponyms, one of them being
the hyponym of the disagreement. The co-hyponym is created through argumentation.
An overlap is resolved by creating a concept that covers the examples at the intersec-
tion of the two overlapping concepts. This new concept causes hypo/hypernymy dis-
agreements where the hypernyms are the two overlapping concepts. When these two
hypo/hypernymies are resolved, the hypernyms are deleted and the overlap is resolved as
well. An indistinguishable disagreement is resolved by deleting one of the two concepts.
Unlike in the other types of disagreement, the examples remain unassigned.
Lexical Disagreements. Lexical disagreements are resolved by changing the signs of
the concepts that cause them. A synonymy disagreement is resolved by changing the
signs of the two synonyms for a new sign. A homonymy disagreement is resolved by
changing the sign of the two homonyms for two different signs.
Untranslatable Disagreements. There is only one concept involved in an untranslatable
disagreement. The sign-intensional definition association of this concept is sent to the
agent that does not have the concept in its contrast set, and this agent will create a new
concept for this association.
Self-Disagreements. A self disagreement can only be caused by an overlap between
two concepts of a same contrast set. Unlike overlap disagreements, however, the self-
disagreements are not solved by creating a new concept but by reassigning some of their
examples. Doing so in self-disagreements does not cause diachronic disagreements, as it
would be the case for overlap disagreements. The agents reassign the examples from the
intersection of these two concepts to these concepts and create new intensional defini-
tions for them through argumentation, creating two new concepts with the same signs as
the old ones.
3.3. Organization of an argumentation
An argumentation over concept meaning goes as follow: first the agents exchange the
sign-intensional definition associations for each of their concepts. Knowing these associ-
ations, each agent can compute the overall pairing relation between each of the concepts
involved. Once done, the agents list their disagreements and proceed to resolve them one
by one. The agents prioritize the resolution of their disagreements in the following or-
der: first self-disagreements, then semantic disagreements, then untranslatable disagree-
ments, and finally lexical disagreements. Each time that a new concept has been created
and added to a contrast set, the agents exchange their new local pairing relations and
compute the missing overall relations. Each time that a new concept has been added or
removed to a contrast set, the agents are updating their list of disagreements. Once there
are no overall pairing relations causing disagreements anymore, the agents update their
lexicon a last time. The termination condition being met, the token is confiscated from
the agents and the argumentation stops.
4. Experiments on Two Agent Disagreements
Our model has multiple properties that can be tested. The first and most important is its
generality, that is to say that our model succeeds to reach mutual intelligibility between
two agents through a monotonic refinement of their contrast sets for all disagreement
types and any combination thereof.
4.1. Experimental Variables
Parameters. There are three parameters in our model: the error threshold τE , the ar-
gument acceptability – a parameter of our inductive learning algorithm ABUI – and the
redundancy of the examples between the two agents’ initial contexts. The error threshold
is set as 5, a value that has been experimentally found challenging in preliminary studies.
The argument acceptability is set at 0.75, the value used in AMAIL tests. The redun-
dancy is set to O%, meaning that the initial contexts of the two agents are always com-
pletely disjoint in our setup. Since the information shared by the two agents is minimal,
this setup is the most challenging for an argumentation.
Set-up Disagreements. The only independent variables of our experiment is the number
and types of disagreement that we set up. Only four types of disagreements are set up:
overlaps, hypo/hypernymies, synonymies and homonymies. A set up disagreement is not
a disagreement as we defined it. A Set-up Disagreement (SD) is a set of inconsistent
right-path associations distributed as training sets among two agents in order to obtain
a specific type of disagreement between the agents once the agents learned their initial
contrast sets as a result of their initial training. SDs are always set-up in the overall
context.
4.1.1. Dependent Variables
In our experiments we measure five dependent variables:
Synchronic Agreement Ratio. The Synchronic Agreement Ratio (SAR) is the ratio be-
tween the of examples from the overall context that are named through a left path associ-
ation with the same unique sign by both agents, over the total number of examples in the
overall context. SAR measures how well the agents have reached mutual intelligibility:
Definition 1 (SAR) Let A1 and A2 be two agents, with contrast sets K1 and K2. The
Synchronic Agreement Ratio of A1 and A2 is:
|{e ∈ UO |e K1l 7→s ∧ e K2l 7→s}|
SAR(A1 , A2 ) =
|UO |
Diachronic Agreement Ratio. The Diachronic Agreement Ratio (DAR) is the inverse of
the diachronic disagreement ratio, a measure which is the ratio between (1) the number
of pairs of examples from an agent’s initial context that were not in a same concept in the
initial contrast set and are in a same concepts the final contrast set, and (2) the number
of pairs of examples that were in two different concepts in the initial contrast set. The
DAR measures how well the principle of concept refinement is maintained throughout
the argumentation process.
Definition 2 (DAR) Let A be an agent that has an initial contrast set K = (U, Q) and a
final contrast set K 0 = (U 0 , Q0 ). The Diachronic Agreement Ratio of A is:
|{e1 , e2 ∈ U|e1 Kl 7→s ∧ e2 Kl 7→s0 ∧ e1 Kl0 7→s00 ∧ e2 Kl0 7→s00 ∧ s 6= s0 }|
DAR(A) = 1 −
|{e1 , e2 ∈ U|e1 Kl 7→s ∧ e2 Kl 7→s0 ∧ s 6= s0 }|
Exchanged Examples Ratio. The exchanged examples ratio (EER) corresponds to the
number of examples that have been sent from one agent to the other through messages,
divided by the number of examples in the overall context.
Coverage Ratio. The coverage ratio (CR) of an agent corresponds to the number of
examples from the overall context that can be associated with a sign by that agent through
left-path associations, divided by the number of examples in the overall context.
Observed Disagreements. The observed disagreements (OD) are proper disagreements
that can be observed between two pairs of agents’ contrast sets at a given time; we mea-
sure their types and number. The OD are different from the SD (that are not proper dis-
agreements). The OD can can be measure either before or after the argumentation.
4.2. Experimental Setup
The generality property of our model is tested over the Soybean data-set. In our experi-
ment, we are only considering the classes that are supported by at least 2 × τE examples,
in order to have enough examples for each agent’s concepts. This leaves 290 examples
distributed over 15 classes. These examples and the sign associated to their class are then
split into two disjoint subsets as right-path associations. The proportion of each sign in
each subset is kept at 1:1. Then, we set-up each type of disagreement by modifying the
sign associated to each example differently in each subset.
Four types of disagreements are being setup in each experiment: overlap, hypo/ hy-
pernymies, synonymies and homonymies. These four types are randomly ordered for
each experiment, and a random number of each disagreement type is selected and then
setup, one type after the other. For instance, if the overlap type has been ordered first,
we select a random number between 0 and the number of classes of Soybean with more
than τE examples divided by three (the number of classes required to setup an overlap
disagreement) and we merge as many times three random classes in order to obtain as
many overlap disagreements. For each of the 200 runs of the experiment, we setup a new
random arrangement of disagreements.
Hypo/hypernyms are set up by replacing a pair of signs by a unique sign in one
of the two subsets. Overlaps are set up by replacing a pair of signs in each subset by a
unique sign, one of the replaced signs being shared by both pairs. Synonyms are set up
by replacing a sign by a new sign in one of the two subsets. Homonyms are set up by
replacing two signs from different subsets by one unique sign.
Once these disagreements have been set up, the two subsets are used by two agent
as a training set to learn its initial contrast set through concept creation using right-path
associations. The examples of a subset become the context of the agent, while the signs of
the subset become its lexicon. After they both learned their initial contrast set, a token is
randomly given to one agent and the argumentation starts. We tested 200 random setups,
followed each time by one argumentation process.
4.3. Experimental Results
Figure 2 shows the results of our experiments measuring SAR, DAR, EER and CR be-
fore and after the argumentation. The left hand plot shows the average of each measure
over the 200 random setups. In the case of DAR and CR, they are also averaged over the
corresponding individual measure of both agents. We can see that the agents significantly
moved toward a synchronic agreement after an argumentation, while diachronic agree-
ment is not compromised. Only a portion of all the examples in the context have been
exchanged, proving that the agents are really reaching mutual intelligibility by argument
exchange, and not by exchanging all of their contexts.The coverage ratio (CR) average
decreases after the argumentation, meaning that less examples can be classified. How-
ever, CR average remains over 0.8, meaning that the global classification of our agents
is not over-fitted on their respective contexts.
The right hand plot in Figure 2 shows the average distance between the individual
measures of SAR, DAR, EER and CR. SAR is calculated locally by using the local
context of each agent instead of the overall context. EER uses the count of the examples
sent by each agent, and CR uses the count of the examples covered by at least one of
each agent’s concepts. Recall that DAR is essentially an individual measure. The scale
on which these distances range goes from 0 to 0.08, meaning that for each measured
ratio, the difference between the two corresponding individual ratios is two orders of
magnitude smaller.
Figure 3 shows the count of observed disagreements at the beginning (left side) and
at the end (right side) of the argumentation, averaged over the 200 different setups. As
expected, new disagreement types have appeared that are not limited to the four types
included in the setups by us. The plots on the top row show the overall disagreements
count, while the bottom row shows the local disagreements count of an individual agent.
Since the argumentation protocol is symmetric and the disagreement setup is random,
the other agent displays a similar profile. We count zero overall disagreements after the
argumentation, showing the general effectiveness of our approach.
Local disagreements may still remain: as local contexts are different from the over-
all context, different pairing relations hold in local contexts than in the overall context.
For instance, local synonymy disagreements are detected after the argumentation. They
are due to a set of concepts that have different signs, but they all have empty adjunct
sets in one local context and therefore are seen as equivalent to one another. Two equiv-
alent concepts that have two different signs cause a synonymy disagreement. If this sit-
uation arises in one local context but not in the other, one agent perceives a local syn-
onymy disagreement while the other agent does not. In the overall context, the two in-
volved concepts each have proper examples in their adjunct sets and therefore they are
not equivalent, so no overall synonymy disagreement arises.
Figure 2. The left plot represents five different types of ratio measured on the overall context before and after
argumentation: synchronic agreement ratio, diachronic agreement ratio, example exchanges ratio and coverage
ratio. Each of these ratio can also be measured on the local context of one agent. The right plot represents,
for each type of ratio presented in the left plot, the distance between the two ratios measured on the two local
contexts of the agents.
Figure 3. The plots show the number of disagreements counted before (left-hand) and after (right hand) the
argumentation, and using (below) the local contexts and (above) the overall context. The x-axis shows the dis-
agreement types labeled as follows: 1. Self-Disagreements, 2. Overlaps, 3. Hypo/hypernymies, 4. Synonymies,
5. Homonymies, 6. Indistinguishable disagreements, 7. Untranslatable disagreements
5. Discussion and Future Work
The experimental results show that our model can significantly improve the mutual intel-
ligibility of two agents having multiple disagreements, while keeping their new shared
classification as a refinement of both old ones. Moreover, the agents do not need to ex-
change an important number of examples from their contexts in order to reach this mu-
tual intelligibility. The mutual intelligibility is attained with low variance, meaning that
the combinations of disagreements that our agents have to effectively resolve does not
impair the ability of our system to reach the mutual intelligibility.
The model used in the experiments uses a systematic strategy to search for and re-
solve every overall disagreements between the agents. The results of another strategy,
a lazy strategy that takes place in a naming game and that triggers an argumentation
only when two agents encounter inconsistent classifications of an example during their
naming game, will be presented in later publications. The hypothesis of generality is not
the only hypothesis tested on our model. Later publications will include tests of: the do-
main independence hypothesis, testing our model over different data sets, the scalability
hypothesis, that our model is able to achieve mutual intelligibility over the contexts of
increasing sizes of a same domain without increasing exponentially the number of ar-
guments exchanged, the simplicity hypothesis, that the number of concepts in the final
contrast sets of our agents are not larger than the number that we could expect from a
brute force alignment after a transfer of all the examples from one agent to another, and
finally the paradigm shift hypothesis, that the SAR and the DAR achieved by an AMAIL
argumentation after an argumentation using our model is better than the SAR and DAR
achieved by an AMAIL argumentation alone, while the total computation time of our
model plus an AMAIL run is lesser than an AMAIL run alone.
Future work will focus on applying our model to different types of classifiers than
inductive learners, most likely deep neural networks. Possible approaches to achieve this
goal have already been discussed in [1].
Acknowledgements
This research has been partially funded by the project ESSENCE: Evolution of Shared
Semantics in Computational Environments (ITN 607062).
References
[1] Adrian, Kemo, From Learning Through Signal Processing to Argumentation on Ontological Represen-
tations, Workshop on Semantic Deep Learning (SemDeep). October 2017.
[2] Adrian, Kemo, and Plaza, Enric, An Approach to Interaction-Based Concept Convergence in Multi-
Agent Systems, The Joint Ontology Workshops Proceedings. May 2017.
[3] Carpenter, B, Typed Feature Structures: An Extension of First-Order Terms. In V.Saraswat & K.Ueda
(Eds.), Logic programming: Proceedings of the 1991 international symposium, 1991, p. 187–201.
[4] Euzenat, J, Interaction-based Ontology Alignment Repair with Expansion and Relaxation, IJCAI, 2017,
p.185-191
[5] Field, H, Logic, Meaning and conceptual role, The Journal of Philosophy, 1977, vol. 74(7), p. 379-409.
[6] Harnard, S, The Symbol Grounding Problem, Physica D: Nonlinear Phenomena, 1990, vol. 42(1-3), p.
335-346.
[7] Laera, L, Blacoe, I, Tamma, V, Payne, R, Euzenat, J, Bench-Capon, T, Argumentation over Ontology
Correspondences in MAS, AAMAS, 2007, p. 228
[8] Ontañón, Santiago, and Plaza, Enric, Coordinated Inductive Learning using Argumentation-based Com-
munication, Autonomous Agents and Multiagent Systems Journal, March 2015, Vol 29, Issue 2 p. 266-
304.
[9] Ontañón, Santiago, and Plaza, Enric, Similarity Measures over Refinement Graphs, Machine Learning
Journal, 2012, 87(1): p.57-92.
[10] Charles Sanders Peirce. Collected Papers. Harvard U.P, 1931-1958.
[11] Rapaport, W, Holism, Conceptual-role Semantics, and Syntactic Semantics, Minds and Machines, 2002,
vol. 12(1), p. 3-59.
[12] Trojahn dos Santos, C, Quaresma, P, Vieira, R, An Argumentation Framework Based on Confidence
Degrees to Combine Ontology Mapping Approaches, IJMSO, 2008, vol. 3(2), p. 142-150