Toward a Civiliza-on of
Prof. Pierre Lévy (Twi;er: @plevy)
Fellow of the Royal Society of Canada
Canada Research Chair in Collec-ve Intelligence
University of O;awa
Evolu-on of Media
(From 2000) Ubiquity, interconnection and animation
of cultural signs (software). Social Computing. New
sign systems. Knowledge economy.
(From 1500) massive technical self-reproduction and
diffusion of the alphabet and other cultural signs. New
languages (animated images, etc.) Scientiﬁc notation
progress. Industrial economy.
(From - 1000) Digitization and universalization of
writing reduced to thirty phonetic signs. Notation of
numbers by position, zero. Commercial economy.
(From - 3000) Autonomous technical memory of
language. Ideographic Signs. Numerals,
measurement units. Agricultural economy.
(From - 300 000) myths, rites, oral transmission,
memory inscribed in matter. Icons. Arts of
memory. Hunting-gathering economy.
Social Media / Social Compu-ng Features
• Global sharing : photo (Flickr), video (Youtube), music / P2P (Bi;orrent),
bookmarks (Delicious), knowledge (Wikipedia, Freebase)
• Distributed crea3on : user‐generated content, blogs (Wordpress), podcasts,
• Social networking : social networks (Facebook, Myspace, Linkedin, Xing,
Pulse, NING...), virtual worlds (Second life), instant micro‐blogging (Twi;er)
• Streaming (Twi;er, Facebook, Friendfeed, Atom or RSS Feeds)
• Mass collabora3on : wikis, opensourcing, crowdsourcing
• Collabora3ve assessment : forums, ra-ngs, reviews (Bazaarvoice)
• Social bookmarking / tagging / categoriza3on (Digg, Delicious, Twine, Diigo,
Stumbleupon, Flickr, YouTube...)
• Cloud compu3ng : data and applica-ons are on‐line in huge distributed
data‐centers (Google, Yahoo, Facebook, Twi;er, Amazon...). SoZware as a
Web n.0, Social Compu-ng, Crowdsourcing : A Global Conversa-on ?
Yes, the digital conversa-on ! But what about seman-c interoperability:
diﬀerent languages, folksonomies, classiﬁca-ons, ontologies...?
Interconnec-on between ideas (via seman-c tags). "
Uniform Seman-c Locator = IEML * concept address **
Collaborative societies of semantic agents, subject-centric computation.
2015 Collective intelligence growth. Augmentation of sense-making.
Interconnec-on between documents (+ data)
Uniform Resource Locator = h;p:// page address.
Centralized search engines, browsers.
1995 Global multimedia public sphere.
Interconnec-on between computers.
Internet Protocol = server address.
Routers, user-friendly PC applications
1980 Personal computing. Virtual communities. "
Digital media convergence.
Interconnec-on between transistors.
Computer memory = bit address. "
Operating systems, programming languages
1950 Augmentation of arithmetical and logical processing.
Computational Collective Intelligence
Seman-c global meta‐ group of
Space IEML / USLs
WWW linked data HTTP / URLs
global meta‐ society of
Internet computer automata
TCP / IP
symbolic pervasive comp.
Compu-ng Chips / OS
manipula-on (mobiles, things,
Augmenta-on Objects in Addressing
of CI rela-on system
Toward a CI Science
scien-ﬁc observatory / digital mirror of CI
Reﬂexive Collec-ve Intelligence
driver of human development
prosperity, health, educa-on, security, peace,
environment, cultural heritages, research, innova-on...
Governance / values
Rights / duties
Messages Equipment / technology
Medias Health / environment
Layers of Seman-c Processing
Points (USLs): 6 primi-ves, sequences of 3L primi-ves (0⩽L⩽6),
categories, catsets, USLs
Perspec-ves: series, matrices, trees
Terms: dic-onary = correspondence points/natural language +
network of seman-c rela-ons between terms
Texts (USLs with a meaning): Gramma-cal rules for the
genera-on of texts automa-cally transl. into nat. languages
Circuits: networks of texts
Flows: circula-on of seman-c energyn in circuits, following the
rules of informa-on economy games
The Bodies of Collec-ve Intelligence
Forms: Sets of USLs and perspec-ves
Colors: Meaning of the USLs
Circuits: graphs of USLs
Flows: economic, electric, neural,... models
3D+t. Addresses Rep of VB User‐controlled
The Nature of Collec-ve Intelligence
group of transforma-on seman-c space
DIGITAL MEDIA form : virtual essence
color : actual essence
(self‐observa-on) symbolic bodies
luminosity : virtual existence
data : actual existence
COGNITION subjec-ve experience : virtual presence
objec-vity : actual presence
3D MATERIAL molecular machines
ECOSYSTEM par-cles / waves
group of transforma-on uniﬁed ﬁeld
Diﬀerences of nature between
IEML and XML / RDF / OWL
• OWL has no seman3c content in itself: no verbs, nouns, adjec-ves,
adverbs, preposi-ons, inﬂec-ons, etc. OWL is rather a ﬁle format for
• IEML has a seman3c content in itself. Users can generate proposi-ons,
complex phrases with several proposi-ons and « texts » from the
syntax and dic-onary of IEML.
• IEML can be expressed in any ﬁle format, including XML, RDF and
OWL. There is currently an automa-c translator from a cursive
nota-on of IEML (called STAR), to binary and XML nota-ons of IEML.
IEML is not a data format!
• It is indeed possible to use IEML to describe OWL ontologies (and so
to decompartmentalize dis-nct ontologies), or to describe in OWL the
complex network of concepts of the IEML dic-onary (like wordnet has
– almost ‐ done for the english language).
Diﬀerences of goals between
IEML and XML / RDF / OWL
• 1) Seman-c interoperability
– Standardizing data formats is already done by the W3C and other standardiza-on
– But diversity of data formats is not the only obstacle to seman-c interoperability :
diversity of ontologies, folksonomies, classiﬁca-on systems, natural languages...
– IEML can be used in the context of ontologies with very diﬀerent hierarchies of
– Once expressed in IEML, a complex concept ‐ the meaning of a *tag ‐ can be
automa-cally translated to any natural language supported by the IEML dic3onary.
• 2) Transparency of seman-c addressing system
• 3) Empowerment of wri-ng / reading
• 4) Symbolic tool for self‐observa-on and self‐reference of collec-ve
These goals have to be addressed by humani-es and social sciences, but
these sciences need the help of soZware engineering.
Toward a transparent
seman-c addressing system (1)
• Opacity by design of the URIs
• By contrast, IEML expressions form a group of
transforma3ons. Automatable algebraic transforma-ons
on IEML symbols correspond to automatable algebraic
transforma-ons on signiﬁca-ons (on “seman-cs”).
• The IEML seman-c space (the immense set of IEML
« texts », called USLs ) is in principle independent of the
URI address space just as it is independent of any physical
or telecommunica-on addressing system.
Toward a transparent
seman-c addressing system (2)
• IEML can bring to the system of URIs a general seman3c
interconnec3on and a full group of transforma3on on
seman3cs. IEML‐URIs can be directly used as concepts in
RDF or OWL. The IEML research program can oﬀer an
alterna-ve grounding to the en--es of the Web of data,
mapping URIs to such IEML‐URIs.
• The power of IEML can be leveraged by the exis-ng
standards of the Web of data. Symmetrically, the
expressive and algebraic proper-es of IEML can leverage
the current Web of data by providing it with a novel
grounding that can make it more seman-c.