=Paper=
{{Paper
|id=Vol-1696/paper4
|storemode=property
|title=Gibsonian Modeling of Users in Social Networks
|pdfUrl=https://ceur-ws.org/Vol-1696/paper4.pdf
|volume=Vol-1696
|authors=Alice Ruggeri,Guido Boella
|dblpUrl=https://dblp.org/rec/conf/lrec/RuggeriB16
}}
==Gibsonian Modeling of Users in Social Networks==
https://ceur-ws.org/Vol-1696/paper4.pdf
Gibsonian Modeling of Users in Social Networks
Alice Ruggeri, Guido Boella
University of Turin, Italy
Center for Cognitive Science, Department of Computer Science
{ruggeri@di.unito.it, boella@di.unito.it}
Abstract
In recent days, social networks are creating huge amounts of data that need to be managed in intelligent ways. Ontologies always
play important roles in these contexts: users and social objects may refer to concepts and this can provide efficient methods to store,
retrieve and recommend contents. Everything, however, relies on the concept of interaction. Contents are produced when people interact
and social objects are used in some way. In the light of this, we present a novel interaction-based ontological approach to deal with
social networks data, managing growth and complexity. In particular, we revisit the standard methodology of computational ontologies
proposing a framework where objects and users are defined as compositions of atomic semantic information, avoiding preventive and
static identification of the system’s players. Our method is inspired by the work of James Gibson, who defined an ecological view of the
human perception based on objects’ natural affordances, in which objects spontaneously give cues about how they can be used depending
on the user who is actually interacting. The idea is that while social objects and users can potentially grow without limits, the spectrum
of all the possible interactions can be the product of limited (and much more simple to represent) links between users and objects’ atomic
semantic information. In this sense, if a user ‘x’ acts on a social object ‘y’, it means that some property of ‘x’ are activated by the action
(i.e., the user embodies a specific role), and some property of ‘y’ makes the action physically possible (i.e., it allows the action to be
performed). In this paper, we show how an interaction-based ontological view can reduce manual efforts while preserving the control
social networks data.
Keywords: Gibsonian Affordances, Ontologies, Social Networks
1. Introduction carries its own functionalities with itself or it inherits them
Social Networks are web-based platforms where users with from a superclass. While this perfectly works in several
different interests and properties interact and live in a kind scenarios, we want to stress the fact that the dynamic part
of virtual second life. Indeed, this has been a real name of the architecture (let us now use terms like actions, inter-
for one of them, SecondLife, where people used to build actions, messages, functionalities, and operators for identi-
avatars, i.e., 3D self-personifications in an invented world. fying this concept) must inhabit inside the objects. In other
The aim of this paper, however, is not to overview all the words, what can happen with an object has to be defined
range of the existing platforms, nor to to make distinctions in the object itself. This somehow freezes the high vari-
and, least of all, not to face social aspects and problematics ability of how an object can be used, and, in general, how
related to the use of such technology. Actually, we want agents can interact with it. Centering Social Networks data,
to focus the attention on how these data coming from users functionalities and visualization directly on users needs the
interactions in social platforms are of scientific interest in rest of the world to be as much flexible as possible from a
terms of knowledge representation and ontology modeling. representational point of view.
We usually refer to the term ontology as a set of formal de- Most of the times, working on objects as main concepts to
scriptions and tools to represent a specific domain (or a part be defined is both practical and sufficient. However, this
of it) in an objective way. This is usually reflected in a defi- strictly depends on the nature of the domain under defini-
nitions of objects with fixed properties, and relations among tion. For example, social networks are extremely dynamic
them that depict the dynamic aspect of the representation. environments where OO-style objects like users, interests,
For instance, we can think at the following description for and locations could be secondary with respect to the inter-
a generic object A: actions that make them active and communicating.
An object A is defined by some attribute p1 , p2 , Our idea regards an ontological modeling of the behavior
and p3 which can have some values within a spe- of intelligent agents, built on top of the concept of affor-
cific numeric range like [1, 10] or among a set of dance introduced by (Gibson, 1977) to describe the process
nominal values, e.g., low, medium, high. Then, underlying the perception. In his work, Gibson (Gibson,
A can exhibit the functionalities f1 and f2 to 1977) claimed that objects assume different meanings de-
the external world, representing its dynamic part, pending on the context, and more specifically, taking into
i.e., its behaviour. account the animal species that interact with them. Im-
plementing this concept in a social network environment
Representing the world by starting from objects and rela- would lead to ontologies based on interactions rather than
tions between them is a classic and intuitive way to make on objects. For instance, let us quickly consider an example
the things working both conceptually and at an application where a user (data creator) publishes on the network some
level. Object-Oriented programming (OOP) is one of the comment about a hole in a specific street. Then, let us as-
most successful programming paradigms that uses this ar- sume we have two types of users (consumers), a cyclist and
chitecture to represent internal data structures. Each object a public-transport passenger, respectively. While the object
under consideration has an high priority for the formers, it RQ #4 how to capture and shape the dynamism and the vari-
can be probably worthless for the latters. In a classic rep- ability of the interactions depending on who/what is
resentation scheme, each object-user combination needs to interacting
be thought and formalized a priori, manually checking all
the possible cases with the relative contraints. Note that this RQ #5 how to enable smart access strategies in dynamic
can be extremely consuming in terms of manual effort (cre- and multimensional data (fuzzy search, graph search,
ation / management of interventions) and carry to unflexi- cross-aspects search, etc.)
ble and redundant representations which do not embed the In general, in social networks, a multitude of combinations
concept of knowledge sharing. In addition to this, not only of aspects must be taken into account depending on several
objects (like street holes) and user types can create a large factors like time, locations, interests, and so on. These are
space of representation, but other dimensions can be added, not well represented by classic paradigms where the world
further multiplying the problem. For instance, the basis is a matter of objects and relationships, since this does not
of the network is the type of interaction, i.e., what people cope with the explosion of cases to define a priori.
can do, creating a three-dimensional space user-interaction-
object which results to be untreatable with classic First Or- 3. Cognitive Background and Related Work
der Logic-like representations (Baldoni et al., 2006). In this section, we overview the main foundations from
Along this contribution, we will talk about ways of think- which our contribution is mostly inspired. Since our pro-
ing at knowledge by means of objective and subjective rep- posal has to do with how ontologies can be used in infor-
resentations, highlighting limits and workarounds. After- mation systems, it is worth to cite important works like
wards, we will present our idea of interaction-based com- (Guarino, 1998)(Gruber, 1995) that deeply describe main
putational ontologies as an approach to solve some of the issues and state-of-the-art approaches. Fiske and Taylor
discussed issues, proposing an implementation to represent (Fiske and Taylor, 2013) highlight important features re-
social networks data. We, finally, conclude the paper with lated to Social Cognition, and computational approaches
a list of future work directions and open problems. are needed to better fit users’ activity.
The starting point of the discussion is the use of formal on-
2. Research Questions tologies. In general, formal ontologies are inspired to the
Social networks have the need of structuring all the data basic principles of the First Order Logic (Smullyan, 1995),
in efficient ways, not only from a computational perspec- where the world is explained by the existence of defined
tive, but rather considering conceptual schemes that better objects and fixed relationships among them. This belongs
enhance the user experience itself. to a physical and static view of the world, since this repre-
In real-life scenarios, it is common to find complex cases sentation is able to treat only the existence of objects and
where data coming from different sources can cross sev- relationships. The same actions are offered to all agents in-
eral aspects, ranging from bureaucracy issues to restaurant teracting with the object, independently of the properties of
reviews. Managing both the quantity and the sparsity of these agents.
the data is the first problem to tackle with advanced tech- Our aim is to manage concepts which have different per-
niques. Then, spreading the data to users according to inter- spectives depending on the kind of agent or species is inter-
ests, actual and current needs, and with the right priority is acting with them, instead of having an object duplicated in
even more challenging. Still, not only social networks usu- different classes according to the different possible behav-
ally have to notify users autonomously, but they also have iors afforded to different agents. A social-driven ontology
to answer to specific user queries. Indeed, the concept of would lie between two extremes, as the first-person ontol-
search in social networks is crucial and partially different ogy mentioned by Searle (Searle, 1998).
from standard information retrival tasks of common search For example, the door provides two different ways to inter-
engines. In fact, the latters have to index data (text, images, act (the set of methods, if we want to use a programming
videos, etc.) to be retrieved by means of classic few-words language terminology): a way for a human user and on the
user queries, whereas, in social networks, queries connect other side the one for a cat. These two ways have some
locations with people, crossing communities, events, and common actions with different implementations, but they
specific time ranges. All this is even made more compli- can also offer additional actions to their agents or players.
cated by the presence of continuously-changing informa- For example, a human can also lock a door with the key or
tion like hashtags, emotional states, and smartphones ap- shut it, whereas a cat cannot do it. The behavioral conse-
plication data. quence of “how to interact with the door” can be “opened
From a computational and ontological perspective, the by the handle” rather than “pushed leaning on it”, and the
challenges faced by this contribution are the following: way the action will be performed is determined by who is
the subject of the action.
RQ #1 how to minimize manual efforts in building computa- The second example has a different character, since it refers
tional ontologies to a technological artifact, i.e., a printer. As such, the ob-
RQ #2 how to represent such complex data maximizing the ject can have more complex behaviours and above all the
sharing of the whole knowledge in a social network behaviours do not depend only on the physical properties
of the agents interacting with it but also with other proper-
RQ #3 how to represent the data without affecting the flexi- ties, like the role they play and thus the authorizations they
bility of objects and agents interactions have. The printer provides two different roles to interact
with it (the set of methods): the role of a normal user, and novel way of estimating kind of affordances at natural lan-
a role of super user. The two roles have some common guage level relying on statistical analysis. Finally, it is
methods (roles are classes) with different implementations, important to refer to (Steedman, 2002), where the authors
but they also offer other different methods to their agents. demonstrated that natural language grammar and planned
For example, normal users can print their documents and actions are related systems.
the number of printable pages is limited to a maximum de- Dynamic taxonomies (Sacco, 2000) exploit a set of in-
termined (the number of pages is counted, and this is a role stances classified in a taxonomy to create latent connections
attribute associated to the agent). between nodes belonging to different paths. In fact, if one
The third example we consider is of a totally different kind. instance is classified under two concepts on different paths
There is no more physical object, since the artifact is an means that there is some link between them that the orig-
institution, i.e., an object of the socially constructed re- inal taxonomy was not aware of. This approach is useful
ality (Searle, 1995). Let us consider a university, where to browse a taxonomy by iteratively selecting nodes for fil-
each person can have different roles like professor, student, tering the data, and in this sense it has some relations with
guardian, and so forth. Each one of these will be associ- every work on making structured knowledge dynamic and
ated to different behaviours and properties: the professors changeable with respect to some context.
teach courses and give marks, and have an income; the stu-
dents give exams, have an id number, and so forth. Here the 4. The Approach
behaviour does not depend anymore on the physical prop- Social networks are a modern way people use to commu-
erties but on the social role of the agent. nicate and share information in general. Facebook, Twitter,
Mental models have been introduced by Laird (Johnson- Flickr and others represent platforms to exchange personal
Laird, 1983), as an attempt to symbolic representations of data like opinions, pictures, thoughts on world-wide facts,
knowledge to make it computable, i.e., executable by com- and related information. All these communities rely on the
puters. This concept is the basis of the most important concept of user profile. A user profile is generally a set of
human-computer cognitive metaphor (Gentner and Stevens, personal information that regard the user in itself as well
2014). his activity within the community.
Another related work which can be considered as a starting Understanding the reference prototype of a user is cen-
point of our analysis is about the link between the Gestalt tral for many operations like information recommendation,
theory (Köhler, 1929; Wertheimer et al., 1927) and the con- user-aware information retrieval, and user modeling-related
cept of affordance in the original way introduced by Gib- tasks. In this context, the concept of affordance can be
son for the perception of objects. Wertheimer, Kohler and used in several scenarios. First, it can be a way to per-
Koffka, the founders of the Gestalt movement (Wertheimer sonalize the content to show to the user according to his
et al., 1927), applied concepts to perception in different interests and activity. This is massively done in today’s
modalities. In particular, it is important to remind the prin- web portals, where advertising is more and more adapted
ciple of complementarity between “figure” and “ground”. to the web consumers. Secondly, the whole content shared
The same concept is applicable in natural language under- by ’user friends’ can be filtered according to his profile, in
standing. For instance, let us think at the sentence “The the same way as in the advertising case. Notice that this
cat opens the door”. In this case, our basic knowledge of does not have to do with privacy issues. In fact, a user may
what the cat is and how it moves can be our ground to un- be not interested in all facts and activities coming from all
derstand the whole figure and to imagine how this action is his friends. Social networks started taking into consider-
performed. In other words, the Gestalt theory helps us say ation these issues, and our proposal regards an ontological
that the tacit knowledge about something (in this case, how modeling of the data that could autonomously and naturally
the cat uses its paws) is shaped on the explicit knowledge of work in this sense.
“what the door is”. Following this perspective, the concepts Commonly, we can think at the interactions in a network as
are not analyzed in a dyadic way, but in a triadic manner. classes managing rules and constraints to match users with
Considering the literature in Object-Oriented programming fixed categories or objects (the terms object and category
(OOP), it is worth citing Powerjava (Baldoni et al., 2006), are interchangeable, referring to “things” that “lives” in the
i.e., an extension of the Java language where an objective network around the users. A scheme of this scenario is il-
and static view of its components is modified and replaced lustrated in Figure 1. Notice that this approach creates one
on the basis of the functional role that objects have inside. class for each combination user-category (when it is seman-
The behavior of a particular object is studied in relation to tically allowed), and it produces a large set of unflexible and
the interaction with a particular user. In fact, when we think predetermined interactions to be formally defined.
at an object, we do it in terms of attributes and methods, For example, if we consider the class StreetHole represent-
referring to the interaction among the objects according to ing the street holes instances in the platform, we need to
public methods and public attributes. The approach is to model all the agents that can interact with it, like CarA-
consider Powerjava-roles as affordances, that is, instances gent (a class modeling the instances of people moving with
that assume different identities dependeing on the agents. cars), BikeAgent, and so forth. The problem is that, with
Weissensteiner and Winter (Weissensteiner and Winter, this methodology, all possible agents and objects have to
2004) focus on landmarks contained in texts to analyze be defined a priori in the ontology, without an appropriate
their role in the general understanding of routes. Distri- uncertainty management. In our approach, we do not look
butional Semantics (Baroni and Lenci, 2010) represents a for a complete coverage of the interacting agents/objects,
Figure 1: Classic view of single interactions connecting Figure 2: In a property-based interaction scheme, cate-
users and categories. Each interaction has its own “life” gories and users disappear from the graph since they be-
and it is different from the others, in the sense that it does come simple compositions of features, while the latters
not share any information nor overlapping degree with the constitute the new basis of the representation. Each interac-
other ones. The distance between the points do not carry tion is thus defined as a set of user/agent properties connect-
any information. ing a set of category/object properties, producing area-like
representations. However, notice that each interaction is ac-
tually formed as multiple and non-contiguous areas, while
since we actually do not represent them as physical con-
the figure has been only created to easily communicate the
cepts, while we only manage sets of fine-grained semantic
concept.
information units that everything (i.e., agents rather than
objects) can have in a specific context/time scenario.
Our idea is illustrated in Figure 2. Objects and users are age, sex, marital status, type of work, location,
substituted by the concept of property (i.e., a semantic in- and a value of affinity for all the objects in the
formation unit), on which interactions directly lie. More in environment (her/his interests)
detail, each interaction is defined as a set of user features
connecting a set of object features, producing area-like rep- In the same way, objects share a set of numeric and nominal
resentations. This way, the need of constructing classes for property O = {o1 , o2 , ..., om }. Examples of them are:
managing all the possible users ∗ actions ∗ objects falls
into an m + n space, where m is the number of user proper- bureaucracy, building, city mantenance, sport,
ties and n is the number of object properties (m and n may education, news, kids, nature, tourism, shopping,
have a certain overlapping degree, however). lost and found, public transport, personal trans-
In this section, we propose a way to model social net- port, hotels, restaurants, culture, entertainment,
works data in a flexible way. As we already anticipated, animals.
in most social networks people can participate in the net-
work through a set of interactions. For instance, some of An example of object vector is the following, representing
them could be the following: a thermal spa in the city centre:
{to buy, to read, to sell, to eat, to drink, to pay object-vector (a thermal spa) public trans-
attention, to work, to learn, to play, to know, to port:0.4, bureaucracy:0.0, building:0.2, city:0.5,
relax, to participate} mantenance:0.0, sport:0.4, education:0.2,
news:0.0, kids:0.2, nature:0.7, tourism:0.5,
Each interaction is defined as two sets of properties or fea- shopping:0.2, lost and found:0.4, personal trans-
tures or semantic information unit, for the users and for the port:0.6, hotels:0.3, restaurants:0.6, culture:0.0,
objects, respectively. A agent/user or a category/object can entertainment:1.0, animals:0.0
be associated to a property with a certain weight. More in
detail, the value for a property can be a value in the range The wights represent a value of how a specific object is re-
[0, 1] representing a degree of affinity within the social net- lated to a property. In the example, a thermal spa results to
work environment, or a nominal value from a given set S be more related to entertainment and transportation rather
(Figure 3). than to bureaucracy and animals. This way, users and ob-
All the users (also called agents and subjects in our jects are defined as vectors in these two multi-dimensional
examples and figures) share a set of properties A = spaces, according to the Vector Space Model (Salton et al.,
{a1 , a2 , ..., an }. Some example of user features are: 1975). Notice that, in this manner, objects that present a
Figure 4: Subjects/users and objects/categories can assume a value in the range [0, 1] for each specific property in the social
network environment. All users and objects are thus represented as vectors. In the same way, a specific interaction becomes
a two-vectors model that represents the association weights with the subjects and the objects properties, respectively. If a
property is not set for a certain user-vector / category-vector / user-interaction-vector / category-interaction-vector, this is
treated as a constraint of having a 0-value for any other vector that will be compared with it (otherwise the total similarity
value between the two vectors will be set to 0).
Scaling (Kruskal, 1964) and Self-Organizing Maps (Koho-
nen, 2001), or by manually-computed ranges.
The first phase concerns the development of the interac-
tion ontology, where the domain experts have to edit a first
sketch (even if this can be tuned by users activities dynam-
ically) of the taxonomy of the interactions. Initially, we
considered a flat organization where interactions work in-
dependently, but the system can work with hierarchy-based
constructions as well. In detail, the knowledge engineer has
to create the two vectors of the model (the one for the user
and the one for the object) for each interaction. An example
is shown later in this section.
At this point, once the interaction ontology with all the
model vectors are created, a user in the network can act
Figure 3: Interaction-centered knowledge representation. according to the adherence between his/her properties with
Interactions are defined by means of two sets of properties, the ones of the existing interactions (their left-side vector
i.e., for the users and for the categories. Users (black nodes in Figure 4), dynamically, and in real-time. The adherence
on the left) and categories (black nodes on the right) satisfy- is computed by means of the well-known cosine similarity.
ing these constraints can participate to the interaction. All From the other side, all the objects are represented as vec-
the interactions are structured in a taxonomy, where sub- tors of features as well. One object can be represented by a
classes inherit both sets of properties from the superclass. value of affinity with all the other objects. This is both prac-
Secondary, users and categories are defined as simple sets tical and plausible, since one object can be related to others
of properties / features. in some way. For example, the category public transport
has a significant degree of affinity with the category private
transport, and it is much higher than what it could be with
similar conceptual nature may change its property status the category sport. There are several ways for computing
and so becoming different things depending on the context. such graded categorizations in automatic ways also in tax-
Each interaction, in the same manner, is defined as two onomy structures as in (Kim and Candan, 2006); however,
vectors of weights (one concerning the user side, and we think that such process must be done manually (or with
one for the objects), and it can be placed within a tax- a manual support), trying to capture the actual semantics
onomy inheriting all the properties from its parents with according to the specific domain of application.
some tuning of the weights. Notice that in case of non- To sum up, the initial modeling efforts lie in the configura-
numerical attributes, the weights can be numerical trans- tion of the interactions by weighting user and object vector
formation obtained by techniques like Multi-Dimensional weights. An example of user-vector model for the interac-
tion to relax is the following: access strategies, since it is the model used for queries and
retrieval by definition (RQ #5).
user-vector age:’any’, location:’any’, sex:’any’,
public transport:0.4, bureaucracy:0.0, build-
ing:0.2, city:0.5, mantenance:0.0, sport:0.7, 6. Conclusions and Future Works
education:0.2, news:0.3, kids:0.2, nature:0.8, In this paper, we proposed an idea for representing the
tourism:0.8, shopping:0.8, lost and found:0.0, knowledge of highly dynamic environments like social net-
personal transport:0.6, hotels:0.4, restau- works and Web Sharing sites. Indeed, these kind of infor-
rants:0.6, culture:1.0, entertainment:1.0, mation need to be carefully organized to remain manage-
animals:0.3 able while making the interaction itself enhanced. We first
started the discussion by thinking at a classic Social Net-
Then, users dynamically change their feature vector
work scenario where users are associated to interests and
through their own activity in the network (and therefore
locations, acting over (virtualized) real-life objects. Then,
they constantly change their interaction scenarios). In ad-
multiple interactions can take place by means of several
dition, as in the basic idea of Dynamic Taxonomies (Sacco,
combinations of these concepts, thus the knowledge com-
2000) by which instances classified under different objects
plexity and the relative management becomes interesting as
are viewed as latent connections between the latters, a real-
much as it gets harder. In future works, we will implement
time adjustment of the weights is not only done by user-
these ideas on real Social Networks data. We advocate an
side, but also on the object vectors. In fact, initial manually-
underlying formalization in first-order logic, in line with
constructed object vectors can exploit the real use carried
flat reification-based approaches such as ((Hobbs, 2008)
by users activities to find unknown affinity connections (or
and (Robaldo, 2011). As pointed out in the introduction,
to moderate the ones already known). This prevents from
a three-dimensional space user-interaction-object results to
incorrect configurations in the cold start.
be untreatable with classic First Order Logic-like represen-
5. Definitions and Validity of the Approach tations (Baldoni et al., 2006), while reification allows to
keep complexity under strict control, thus providing a scal-
The entities involved in our proposal are the following: able instrument to implement our model. Then, we will
• Property. Also called semantic information unit, integrate this approach with automatic techniques to ex-
it represents the central brick of the world under tract, recommend and visualize interaction-based and user-
representation. Every agent/object/interaction is built centered contents by using the proposed ontology mod-
on top of it. eling, relying on semantic technologies and visualization
tools such as (Di Caro et al., 2011; Boella et al., 2014; Can-
dan et al., 2012; Cataldi et al., 2013).
• Agent/Object. It is a set of pairs < p, v > where
p is a property and v is a value within its domain D(p). 7. Bibliographical References
Baldoni, M., Boella, G., and Van Der Torre, L. (2006).
• Interaction. An interaction is a pair of left and right powerjava: ontologically founded roles in object ori-
property sets, defining who interacts with what. ented programming languages. In Proceedings of the
2006 ACM symposium on Applied computing, pages
1414–1418. ACM.
• Interaction Taxonomy. Interactions are organized in Baroni, M. and Lenci, A. (2010). Distributional memory:
a taxonomical structure such that if an interaction Ip A general framework for corpus-based semantics. Com-
is parent of an interaction Ic , then all left and right putational Linguistics, 36(4):673–721.
property sets of Ip are inherited by Ic .
Boella, G., Di Caro, L., Ruggeri, A., and Robaldo, L.
(2014). Learning from syntax generalizations for auto-
Agents and objects are compositions of properties, so there matic semantic annotation. Journal of Intelligent Infor-
is no need to build user- and object ontologies. This mini- mation Systems, 43(2):231–246.
mizes manual efforts in building computational ontologies Candan, K. S., Di Caro, L., and Sapino, M. L. (2012). Phc:
(see Research Question (RQ) #1 in Section 2.) exploiting Multiresolution visualization and exploration of text cor-
the efficacy of the vectorial representations. In the same pora with parallel hierarchical coordinates. ACM Trans-
way, the model maximizes the sharing of knowledge since actions on Intelligent Systems and Technology (TIST),
objects and agents use the same feature space (RQ #2). 3(2):22.
Then, the flexibility of the interactions is not affected by Cataldi, M., Caro, L. D., and Schifanella, C. (2013). Per-
such representation (RQ #3) since they directly rely on sonalized emerging topic detection based on a term ag-
them by being modeled in the same fashion by two fea- ing model. ACM Transactions on Intelligent Systems and
ture vectors. In addition, the actions that the platform can Technology (TIST), 5(1):7.
take can be easily defined with constraints on the agent- Di Caro, L., Candan, K. S., and Sapino, M. L. (2011). Nav-
and object vectors, so it is possible to shape the behaviour igating within news collections using tag-flakes. Journal
of the social network (for the same interaction) depending of Visual Languages & Computing, 22(2):120–139.
on who/what is interacting (RQ #4). Finally, the use of nu- Fiske, S. T. and Taylor, S. E. (2013). Social cognition:
merical vectors completely fits the requirements of smart From brains to culture. Sage.
Gentner, D. and Stevens, A. L. (2014). Mental models.
Psychology Press.
Gibson, J. J., (1977). The Theory of Affordances. Lawrence
Erlbaum.
Gruber, T. R. (1995). Toward principles for the design of
ontologies used for knowledge sharing? International
journal of human-computer studies, 43(5):907–928.
Guarino, N. (1998). Formal Onthology in Information
Systems: Proceedings of the First International Confer-
ence (FIOS’98), June 6-8, Trento, Italy, volume 46. IOS
press.
Hobbs, J. R. (2008). Deep lexical semantics. In Proc. of
the 9th International Conference on Intelligent Text Pro-
cessing and Computational Linguistics (CICLing-2008),
Haifa, Israel.
Johnson-Laird, P. (1983). Mental models: Towards a cog-
nitive science of language, inference, and consciousness.
Number 6. Harvard University Press.
Kim, J. W. and Candan, K. (2006). Cp/cv: concept similar-
ity mining without frequency information from domain
describing taxonomies. In Proceedings of the 15th ACM
international conference on Information and knowledge
management, pages 483–492. ACM.
Köhler, W. (1929). Gestalt psychology.
Kohonen, T. (2001). Self-organizing maps, volume 30.
Springer.
Kruskal, J. B. (1964). Multidimensional scaling by opti-
mizing goodness of fit to a nonmetric hypothesis. Psy-
chometrika, 29(1):1–27.
Robaldo, L. (2011). Distributivity, collectivity, and cu-
mulativity in terms of (in)dependence and maximal-
ity. The Journal of Logic, Language, and Information,
20(2):233–271.
Sacco, G. (2000). Dynamic taxonomies: A model for large
information bases. Knowledge and Data Engineering,
IEEE Transactions on, 12(3):468–479.
Salton, G., Wong, A., and Yang, C. S. (1975). A vector
space model for automatic indexing. Commun. ACM,
18(11):613–620, November.
Searle, J. R. (1995). The construction of social reality. Si-
mon and Schuster.
Searle, J. (1998). Mind, language and society: Philosophy
in the real world. Cambridge Univ Press.
Smullyan, R. (1995). First-order logic. Dover Publica-
tions.
Steedman, M. (2002). Plans, affordances, and combina-
tory grammar. Linguistics and Philosophy, 25(5):723–
753.
Weissensteiner, E. and Winter, S. (2004). Landmarks in
the communication of route directions. Geographic in-
formation science, pages 313–326.
Wertheimer, M., Köhler, W., and Koffka, K. (1927).
Gestaltpsychologie. Einfuhrung in die neuere Psycholo-
gie. AW Zickfeldt, Osterwieck am Harz.