Transcript Title of the presentation

Social Networks: from communication to solidarity (an interdisciplinary approach)
Fundación Sierra-Pambley, León (Spain)
Understanding complexity: systems, emergence and evolution
León, January 28 2014
Participating and
Anticipating
Actors and Agent Networks.
Social Computing
Gordana Dodig Crnkovic
Professor of Computer Science
Mälardalen University,
School of Innovation, Design and
Engineering
[email protected]
Mälardalen University, Sweden
Abstract
Computing is becoming ubiquitous and essential part of
human society. As participants in a major technological and
cultural change caused by ICT, we want to be able to
understand ongoing processes and anticipate future
possibilities. When we introduce changes in the society we
want to know the possible outcomes. That is the goal of social
computing.
Very often programs of social change envisage certain goals,
but it is not at all clear how they can be reached. Unfavorable
development can result from lack of understanding of possible
consequences of interventions, decisions and other changes
in
social
systems.
p. 3
Abstract
As social agents we have intuitions on the interpersonal level
with focus on individual people. Computational techniques
can augment our understanding on a social level, where we
have limited everyday experiences.
In this talk I will present an example of the analysis of city as a
socio-computational system.
p. 4
Two aspects of social computing
One important factor in human project of development of prosperous global
society is understanding of the behavior of social systems. Computing as a
method provides means for this study in a form of computational models
and
simulations
that
are
being
developed.
There are two different types of social computing (Wang et al. 2007),
centered on its two different aspects:
1. computing mechanisms and principles with computational modeling of
groups of agents exchanging information in networks and
2. human aspects of social computing (critical theory), with social side of
social web applications such as blogs, wikis, social bookmarking, instant
messaging, and social networking sites and crowdsourcing.
This lecture focus on computational aspects of social computing and its
relation to models of computing as information processing.
p. 5
Computer as a communication device
Even though computers were invented in order to
automatize calculations [Hilbert program (1900); Turing
Machine (1936)], after a while the importance of the
computer as a communication device was recognized, with
its important consequent shared knowledge and
community-building (Licklider and Taylor 1968).
Hilbert, David, 1900, “Mathematische Probleme”, Nachrichten von der Königlichen
Gesellschaft der Wissenschaften zu Göttingen, Math.-Phys. Klasse, 253-297. Lecture
given at the International Congress of Mathematicians, Paris, 1900.
Turing AM. On Computable Numbers, with an Application to the
EntscheidungsproblemÊ. Proceedings of the London Mathematical Society, series 2,
1936; 42:230-265.
Licklider, J.C.R. and Taylor R. W. (1968) The computer as a communication device.
Science and Technology (September), 20-41.
p. 6
From information communication to
social intelligence
Social computing with the focus on social is a phenomenon
which enables extended social cognition,
while the Social computing with the focus on computing is
about computational modelling and it is a new paradigm of
computing used for understanding underlying mechanisms.
p. 7
Conceptual Basis: Network Modells
Evolutionary network
Human neural network
Social network
Protein interaction network in human cell
8
Human groups are information processing
networks and knowledge generators
http://www.google.com/insidesearch/features/search/knowledge.html
Google Knowledge Graph
http://press.emerson.edu/imc/files/2011/12/social-network.jpg
9
Networks of networks of information
– complexity
In a complex system, what we see is dependent on where we are and what sort of
interaction we use to study the system.
Computational approaches to
complex systems can be descriptive
or generative models
Generative models answer the
question: How does the complexity
(emergent properties) arise?
Evolution is the most well known
generative mechanism for
generating increasingly complex
systems (organisms).
http://www.morphwize.com/company/index.php?option=com_k2&view=itemlist&task=tag&tag=complex+system+solution
p. 10
Complex systems definition
“ … systems that comprise many interacting parts with the
ability to generate a new quality of collective behaviour
through self-organization, e. g. the spontaneous formation of
temporal, spatial or functional structures. They are therefore
adaptive as they evolve and may contain self-driving feedback
loops. Thus, complex systems are much more than a sum of
their parts. Complex systems are often characterized as having
extreme sensitivity to initial conditions as well as emergent
behaviours that are not readily predictable or even completely
deterministic.”
Robert Meyers, Encyclopaedia of Complexity and Systems Science
(2009)
p. 11
Social Networks
Social networks (SN) are social structures made of nodes
(which are, generally, individuals or organizations) that are
tied by one or more specific types of interdependency, such as
values, visions, ideas, financial exchange, friends, kinship,
dislike, conflict, trade, web links, disease transmission, etc.
12
Social Networks
Social networks analysis plays an important role in studying
the way specific problems of social groups are solved
(organizational, economic, ecological, medical,
epidemiological etc.) and the ways in which individuals within
socials groups succeed in achieving their goals.
Social networks analysis has addressed also the dynamics
issue, called dynamic networks analysis. This is an emergent
research field that brings together traditional social network
analysis, link analysis and multi-agent systems.
13
Study of networks regularities:
Laszlo Barabasi
See the work of Albert-László Barabási who studies networks on different scales.
http://www.barabasilab.com/pubs-talks.php
http://www.youtube.com/watch?v=10oQMHadGos
14
Characteristics of complex networks
Self-similarity of complex networks1
Complex networks have been studied extensively owing to
their relevance to many real systems such as the world-wide
web, the Internet, energy landscapes and biological and social
networks. A large number of real networks are referred to as
‘scale-free’ because they show a power-law distribution of the
number of links per node.
However, it is widely believed that complex networks are not
invariant or self-similar under a length-scale transformation.
This conclusion originates from the ‘small-world’ property of
these networks, which implies that the number of nodes
increases exponentially with the ‘diameter’ of the network,
rather than the power-law relation expected for a self-similar
structure.
1Chaoming
Song, Shlomo Havlin & Hernán A. Makse
p. 15
Self-similarity of complex networks
Chaoming Song, Shlomo Havlin & Hernán A. Makse
Here we analyse a variety of real complex networks and find
that, on the contrary, they consist of self-repeating patterns
on all length scales. This result is achieved by the application
of a renormalization procedure that coarse-grains the system
into boxes containing nodes within a given ‘size’. We identify a
power-law relation between the number of boxes needed to
cover the network and the size of the box, defining a finite
self-similar exponent.
These fundamental properties help to explain the scale-free
nature of complex networks and suggest a common selforganization dynamics.
http://www.nature.com/nature/journal/v433/n7024/full/nature03248.html Self-similarity of
complex networks Nature 433, 392-395 (2005)
p. 16
Origins of fractality in the growth of
complex networks
Chaoming Song, Shlomo Havlin & Hernán A. Makse
Complex networks from such different fields as biology,
technology or sociology share similar organization principles.
The possibility of a unique growth mechanism promises to
uncover universal origins of collective behaviour. In particular,
the emergence of self-similarity in complex networks raises
the fundamental question of the growth process according to
which these structures evolve. Here we investigate the
concept of renormalization as a mechanism for the growth of
fractal and non-fractal modular networks.
http://www.nature.com/nphys/journal/v2/n4/full/nphys266.html
Nature Physics 2, 275 - 281 (2006)
p. 17
Origins of fractality in the growth of
complex networks
Chaoming Song, Shlomo Havlin & Hernán A. Makse
Song, Havlin and Makse investigate the concept of
renormalization as a mechanism for the growth of fractal and
non-fractal modular networks. They show that the key
principle that gives rise to the fractal architecture of networks
is a strong effective ‘repulsion’ (or, disassortativity) between
the most connected nodes (that is, the hubs) on all length
scales, rendering them very dispersed.
A robust network comprising functional modules, such as a
cellular network, necessitates a fractal topology, suggestive of
an evolutionary drive for their existence.
http://www.nature.com/nphys/journal/v2/n4/full/nphys266.html
Nature Physics 2, 275 - 281 (2006)
p. 18
Generative approaches. Agent-based
Models (ABM)
An agent-based model (ABM) is a computational model for
simulating the actions and interactions of autonomous
individuals in a network, with a view to assessing their effects
on the system as a whole. They are used in study of
complexity and emergence.
It combines elements of game theory, complex systems,
emergence, computational sociology and evolutionary
programming. Monte Carlo Methods are used to introduce
randomness.
http://www.youtube.com/watch?v=2C2h-vfdYxQ&feature=related Composite Agents
http://en.wikipedia.org/wiki/Agent-based_model
p. 19
Agent-based models of dynamic
adaptive systems
ABMs in general are used to model complex, dynamical
adaptive systems. The interesting aspect in ABMs is the micromacro link (agent-society). Multi-Agent Systems (MAS) models
may be used for any number (in general heterogeneous)
entities spatially separated by the environment which can be
modeled explicitly.
Interactions are in general asynchronous which adds to the
realism of simulation.
Social computing represents a new computing paradigm
which is one sort of the natural computing, often inspired by
biological systems (e.g. swarms).
p. 20
Simulation
The main tools in this field are simulation techniques used in
order to facilitate the study of society and to support decisionmaking policies, helping to analyze how changing policies
affect social, political, and cultural behavior (Epstein, 2007).
Epstein, J. M. (2007). Generative Social Science: Studies in Agent-Based Computational
Modeling. Princeton University.
p. 21
Emergence of global social computing
With regard the human aspect, social computing is radically
changing the character of human relationships worldwide
(Riedl, 2011). Instead of maximum 150 connections prior to
ICT, (Dunbar, 1998), social computing easily leads to networks
of several hundred of contacts.
It remains to understand what type of society will emerge
from such massive “long-range” distributed interactions
instead of traditional fewer and deeper short-range ones.
Riedl J. (2011) "The Promise and Peril of Social Computing”, Computer, vol.44, no.1, pp. 93-95
Dunbar R. (1998) Grooming, Gossip, and the Evolution of Language, Harvard Univ. Press
p. 22
Towards social intelligence
In this process, information overload on individuals is steadily
increasing, and social computing technologies are moving
beyond simple social information communication toward
social intelligence, (Zhang et al. 2011) (Lim et al. 2008) (Wang
et al. 2007), which brings an additional level of complexity.
Zhang D., Guo B., Yu Z. (2011) Social and Community Intelligence, Computer, Vol. 99, No.
PrePrints. doi:10.1109/MC.2011.65
Lim H. C., Stocker R., Larkin H. (2008) Ethical Trust and Social Moral Norms Simulation: A
Bio-inspired Agent-Based Modelling Approach. In: 2008 IEEE/WIC/ACM International
Conference on Web Intelligence and Intelligent Agent Technology, December 2008. pp.
245-251.
Wang F-Y., Carley K. M., Zeng D., and Mao W. (2007) Social Computing: From Social
Informatics to Social Intelligence. IEEE Intelligent Systems 22, 2 (March 2007), 79-83.
DOI=10.1109/MIS.2007.41 http://dx.doi.org/10.1109/MIS.2007.41
p. 23
The emergence of social institutions
from individual interactions
As Gilbert (2005) points out, novelty of agent based models
(ABMs) “offers the possibility of creating ‘artificial’ societies in
which individuals and collective actors such as organizations
could be directly represented and the effect of their
interactions observed. This provides for the first time the
possibility of using experimental methods with social
phenomena, or at least with their computer representations;
of directly studying the emergence of social institutions from
individual interaction.”
Gilbert N: (2005) Agent-based social simulation: dealing with complexity,
http://www.complexityscience.org/NoE/ABSS-dealing%20with%20complexity-1–1.pdf
p. 24
Crowdsourcing
Crowdsourcing is, according to the Merriam-Webster
Dictionary, the practice of obtaining needed services, ideas, or
content by obtaining contributions from a large group of
people, and especially from an online community, rather than
from traditional employees or suppliers.
Tools such as prediction markets, social tagging, reputation
and trust systems as well as recommender systems are based
on crowdsourcing. Polymath Project is an example of
mathematical proofs made by crowdsourcing
http://polymathprojects.org/
http://www.nextscientist.com/3-examples-crowdsourcing-science/
25
Computational modelling of social
behavior
Another focus of social computing is on computational
modeling of social behavior, among others through Multiagent systems (MAS) and Social Networks (SN).
There are several usages of Multi-agent systems: to design
distributed and/or hybrid systems; to develop philosophical
theory; to understand concrete social facts, or to answer
concrete social issues via modelling and simulation.
26
Multi-agent systems for modelling of
social behavior
Multi-agent systems are used for modelling, among other
things, cognitive or reactive agents who interact in dynamic
environments where they possibly depend on each other to
achieve their goals.
The emphasis is nowadays on constructing complex
computational systems composed by agents which are
regulated by various types of norms, and behave like human
social systems.
27
An Interlude:
Information, computation,
cognition Agency-based Hierarchies of Levels
Human connectome
http://outlook.wustl.edu/2013/jun/human-connectome-project
scientificamerican0612-50.pdf The Human Brain Project
http://www.nature.com/scientificamerican/journal/v306/n6/pdf/
information
computation
cognition
A framework based on
an agents perspective.
p. 28
http://www.idt.mdh.se/~gdc/work/PT-AI-Oxford-2013.09.21-GDC.pdf
Information, computation, cognition.
Agency-based Hierarchies of Levels
Short summary of the Info-computational framework:
1.
Information constitutes a structure consisting of
differences in one system that cause the differences in
another system. In other words, information is
<observer>-relative.
2.
Computation is information processing (dynamics of
information). It is physical process of morphological
change in the informational structure (physical
implementation of information, as there is no information
without physical implementation.)
p. 29
Information, computation, cognition.
Agency-based Hierarchies of Levels
3.
Both information and computation appear on many
different levels of
organisation/abstraction/resolution/granularity of
matter/energy in space/time.
4.
Of all agents (entities capable of acting on their own
behalf) only living agents have the ability to actively
make choices so to increase the probability of their own
continuing existence. This ability of living agents to act
autonomously on its own behalf is based on the use of
energy and information from the environment.
p. 30
Information, computation, cognition.
Agency-based Hierarchies of Levels
5. Cognition consists of all (info-computational) processes
necessary to keep living agent’s organizational integrity on
all different levels of its existence.
Cognition = info-computation
6. Cognition is equivalent with the (process of) life.
Its complexity increases with evolution.
This complexification is a result of morphological
computation.
http://www.idt.mdh.se/~gdc/work/Information-Computation-Agency-Cognition-20131218.pdf
p. 31
City as Complex System
As urban planning moves from a centralized,
top-down approach to a decentralized,
bottom-up perspective, our conception of
urban systems is changing. Michael Batty
studies urban dynamics in the context of
complexity theory, presenting models that
demonstrate how complexity theory can
combine a huge number of processes and
elements into organic wholes. The bottom-up
processes -- in which the outcomes are
always uncertain -- can combine with new
forms of geometry associated with fractal
patterns and chaotic dynamics to provide
theories that are applicable to highly complex
systems such as cities.
http://www.complexcity.info/files/2011/07/batty-cluster-magazine-2008.pdf Generating cities bottom up
32
City as Complex System
http://www.complexcity.info
http://www.complexcity.info/ A Science of Cities
Models based on cellular automata (CA),
simulate urban dynamics through the
local actions of automata. Agent-based
models (ABM), in which agents are
mobile and move between locations
relate to many scales, from the scale of
the street to patterns and structure at the
scale of the urban region. Applications of
all these models are used to specific
urban situations, discussing concepts of
criticality, threshold, surprise, novelty,
and phase transition in the context of
spatial developments.
33
A New Model for Urban Scaling – explaining
how quantities scale with population size. Luis Bettencourt
The article “The Origins of Scaling in Cities” which appeared in
June 2013 issue of Science has a simple mathematical model
of a city: it is a combination of a social network of interactions
and something actually embedded in physical space, that is
what a city is—an agglomeration of people in a specific
region.
Even though the model is simple, it works surprisingly well.
Cities are modeled as massive social networks, made not so
much of individual characteristics of people but of their
interactions. These social interactions happen, inside other
networks – social, spatial, and infrastructural – which together
allow people, things, and information to meet across urban
space.
34
Four simple assumptions of
Bettencourt’s model
(1) Mixing population. The city develops so that citizens can explore it
fully given the resources at their disposal.
(2) Incremental network growth. This assumption requires that
infrastructure networks develop gradually to connect people as they
join, leading to decentralized networks
(3) Human effort is bounded … [as a city grows, our ability to deal with
it can't increase beyond some reasonable amount]
(4) Socioeconomic outputs are proportional to local social interactions
… From this perspective, cities are concentrations not just of people,
but rather of social interactions.
http://www.santafe.edu/news/item/science-bettencourt-cities-framework/
http://www.wired.com/wiredscience/2013/06/a-new-model-for-urban-scaling
35
A word of caution:
Using computational models
Computational models are excellent tools for analysis of social
phenomena as they give us an overview and possibility to
study different possible scenarios in the development of social
systems. However, they cannot replace good human
judgment.
All simulations depend on the assumptions of the model, so in
other words the old wisdom holds: garbage in – garbage out!
With this in mind, social computing can be expected to
develop into everyday tools for decision-making concerning
social phenomena – in economics, government, business,
medicine, bioinformatics, epidemiology, politics, urban
planning, etc.
36
Further Reading: A Computable Universe
Computational models and methods can be used not only to study
social phenomena, but they can be applied to whole of nature too.
Here is some new research on this bigger topic.
p. 37
Computing Nature
Computation, Information, Cognition
Information and Computation
Computing Nature
Editor(s): Gordana Dodig Crnkovic and Susan
Editor(s): Gordana Dodig Crnkovic and
Editor(s): Gordana Dodig Crnkovic and
Stuart, Cambridge Scholars Publishing, 2007
Mark Burgin, World Scientific, 2011
Raffaela Giovagnoli, Springer, 2013
http://astore.amazon.co.uk/books-books21/detail/9814295477
http://dx.doi.org/10.1007/978-3-64237225-4
http://www.amazon.co.uk/ComputationInformation-Cognition-NexusLiminal/dp/1847180906/ref=sr_1_2?ie=UTF
8&qid=1306954122&sr=12#reader_1847180906
p. 38
This talk is based on the following work
Dodig-Crnkovic G., Dynamics of Information as Natural Computation, Information 2011, 2(3),
460-477; doi:10.3390/info2030460 Special issue: Selected Papers from FIS 2010 Beijing
Conference, 2011.
http://www.mdpi.com/journal/information/special_issues/selectedpap_beijing
http://www.mdpi.com/2078-2489/2/3/460/ See also:
http://livingbooksaboutlife.org/books/Energy_Connections
Dodig-Crnkovic, G.; Rotolo, A.; Sartor, G.; Simon, J. and Smith C. (Editors)
Social Computing, Social Cognition. Social Network and Multiagent Systems. Social Turn SNAMAS 2012
AISB/IACAP World Congress 2012. Birmingham, UK, 2-6 July
2012http://events.cs.bham.ac.uk/turing12/proceedings/11.pdf , 2012.
Dodig-Crnkovic G., Agent Based Modeling with Applications to Social Computing, Proceedings
IACAP 2011. The Computational Turn: Past, Presents, Futures?, p 305, Mv-Wissenschaft,
Münster, Århus University, Danmark, Editor(s): Charles Ess and Ruth Hagengruber, July 2011.
http://www.idt.mdh.se/~gdc/work/IACAP11_Gordana-DC.pdf
Dodig-Crnkovic, G. Information, Computation, Cognition. Agency-based Hierarchies of Levels. In
V. C. Müller (Ed.), Fundamental Issues of Artificial Intelligence (Synthese Library). Berlin: Springer.
(forthcoming) http://www.idt.mdh.se/~gdc/work/Information-Computation-Agency-Cognition20131218.pdf
p. 39