Mozaika The Humanizing Technologies Lab
Mariana Damova, PhD 13.07.2015 Montreal
Data Science - Use Cases
The meaning …
1. Humanizing technologies - technologies that humanize (their users)
2. Humanizing technologies - technologies that are human (close to the human)
3. Humanizing technologies - technologies that enable human to better use ICT
human-centric technology
How technology works …
The phrase “humanizing emerging technologies” is about reducing the amount of mystery around how a technology works and about helping people retain a sense of control over their changing environments.
http://radar.oreilly.com/2014/04/humanizing-emerging-technologies.html
Big Data, Linked Data and The Clouds
Volume Variety
Velocity
- Structured - Unstructured - Semi-structured - All of the above
- Terabytes - Records - Transactions - Tables, files
- Batch - Near time - Real time - Stream
Natural Interfaces
• Gesture control https://www.youtube.com/watch?v=91aDt0UHcUo
• Wearables • Google Glass
• Samsung watch
• Voice biometrics • Voice control • Dialogue
https://www.youtube.com/watch?v=MpjpVAB06O4
Nuance Voice Biometrics
IVR Authentication Use Case
“At VB Bank, my voice is my password”
Carol Foster
ID VERIFIED a
Slide credit: Nuance Communications
Virtual Personal Assistant
2010 Denise - https://www.youtube.com/watch?v=7W52TL9Akv4
2012 Denise - https://www.youtube.com/watch?v=LQkpobxbHTY
2014 Google Now vs. Siri vs. Cortana - http://www.cnet.com/how-to/google-now-vs-siri-virtual-assistants-duke-it-out-video/ - http://www.cnet.com/news/cortana-vs-siri-vs-google-now/ Use for search: Google Search, Wolfram Alpha, Bing Siri uses Nuance for speech recognition
empathy
In the future insight
trust
reliance
actionable knowledge
Better Human Understanding,
not Big data
is the Future of Business.
That is where Mozaika is going
Mozaika is an SME and a Research Center
- semantic data mining, natural language processing, human-computer interaction, data science
- information infrastructures serving variety of applications such as enhancing creativity
- cultural heritage cataloguing, smart cities
- consulting in project development
- etc.
Implied technologies
Semantic Web Technologies - Breaking the data siloes
Linked Open Data Cloud
Linked Open Vocabularies
FactForge
Natural Language and the Semantic Web
DBpedia URI
@Davidcamposh has visto el de Una verdad incomoda de <Al Gore>...es muy bueno tambi Davidcamposh’ve seen An Inconvenient Truth of <Al Gore> ... is very good also
positive sentiment topic: Al Gore
Person DBpedia URI
Politician
United States
hasProfession
bornIn
EN FR DE
Who painted Mona Lisa? Qui a paint Mona Lisa? Wer hat Mona Lisa gemahlt?
Who is Mona Lisa’s painter? Qui est le paintre de Mona Lisa? Wer ist der Mahler von Mona Lisa ?
Who created Mona Lisa? Qui a créé Mona Lisa? Wer hat Mona Lisa geschöpft?
Different language Different syntax Different lexicon
Same semantics
RDF
:Painter :Painting
:painted
Mona Lisa ? rdf:typ
e rd
f:ty
pe
Leonardo Mona Lisa
RDF Repository
SPARQL
Missing Piece A Multilingual SPARQL-Based Retrieval Interface for Cultural Heritage Objects
Reason-able View Linked Open Data
from the Cultural Heritage domain Gothenburg City Museum
Europeana, DBPedia, CIDOC-CRM
SPARQL End-point
Coverage: 1159 query patterns in 15 languages: Bulgarian, Finish, Norwegian, Catalan, French, Romania, Danish, Hebrew, Russian, Dutch, Italian, Spanish, English, German, Swedish 10 characteristics of cultural heritage objects: creation date, time period, material, title, dimension, current location (museum and city), color, author, type
Evaluation Random queries in 7 languages with very few native informants corrections
Extendibility Writing a new query grammar requires 150 lines of code
Linguistic Linked Open Data Cloud
Multilingual Single Digital Market
• Break the language blocking – Single languages address no more than 20% of the Digital Single
Market
• Enable seamless use of all official languages of the EU – Ensure open access to over 50% of the world’s online
population and 73% of the world online market – Approximately 60% of individuals in non-Anglophone countries
seldom or never make online purchases from English-language sites
• Language technology made in Europe – will transform Europe into a world-wide leader in technology
innovation – Will secure Europe’s future as a world-wide trader and
exporter of goods, services and information
6/4/2015 EuroDIG 2015
META and LT-Innovate
Multilingual Europe: The Crowning Touch to the Digital Single Market
Mozaika’s Current and Past projects …
Human Resources Management
• Semantic representation of skills, competences, geographical information industries relatedness, organizational and personal information • Semantic matching
with ProfiCV
From module of the ProfiCV system towards DaaS
CITYSUMMARIES
with Digital Spaces Living Lab
Real time multi-modal summarization of city experiences and information - trip planning - while visiting - trip memories catcher
mobile and web-based
Smart Cities application
Information Management for Cosmic Studies
Small Communication Satellite Mission Space Technology and Research Insitute at the Bulgarian Academy of Sciences
Sofia State University University “Kliment of Ohrid”
RaySat Ltd. (satellite networks ompany)
• Support and enhancement of science research and human activities in Antarctica
• Hi-speed two-way backhaul data transfer for scientific, safety and other applications
• Off-line two-way operational communication services for professional personal or rescue purposes
• Continental surface measurements of biological and natural phenomena
• Weather monitoring and forecasting
Communications project, aiming to mobilize scientific and industrial effort to build a purely Bulgarian product with international impact.
10-60 Mbps data-transfer bit rate
1500-600-km orbit altitude
remote sensing of Earth exploration
with
Geo Linked Data
Interactive map of the Bulgarian Dialects
with IBL - BAS
http://ibl.bas.bg//bulgarian_dialects/
DM2E – Digital Manuscripts to Europeana
Codex Suppraliensis
metadata in DM2E format
http://csup.ilit.bas.bg/node/1
with ILI – BAS and DM2E project
http://dm2e.eu/
Virtual itineraries
http://bulgarianheritage.bulgariana.eu/jspui/handle/pub/624/browse?type=name&submit_browse=%D0%9E%D0%B1%D0%B5%D0%BA%D1%82%D0%B8
Idea initiated from 3D laser scanned architectural objects and an undergraduate course in Multimedia at NBU
The shown items are made with Europeana and Geocad93. They are part of Bulgariana collection and are currently hosted at Ontotext.
starting with Sofia Holy Forest
Publishing
• Linked Open Data publishing • E-Publishing, E-books • Intelligent reading and writing assistants • Scientific literature publishing
with Springer Verlag with Sofia University and others
Text and Image
- Association based natural language processing - Sentiment expression - Lexical semantics - Visual lexicon
Capturing human and personal characteristics based on the tags chosen for an image
So, it will be …
These are the Humanizing Technologies
Image Credit: Robin Bertolletti