1 CELI – Language and Information Gennaio 2014
2 We develop software solutions based on (NLP) Natural Language Processing
3 CELI’s offices, Countries in which we operate, Years of experience, People, Active customers, Business lines
4 Partners in Academia, Research projects, Published scientific papers
Close relationship with scientific community
5 From 1999 to 2013
6 Clients: semantic solutions, Speech Technology, Blogmeter
7 NLP solutions
8 NLP technology: Comprehensive suite of multilingual components and resource
9 Linguistic processing and annotation
10 From text to Knowledge
11 Meaningful intelligence from unstructured information
12 Speech technology: Comprehensive suite of multilingual components and resources for text processing in Voice application (Text To Speech)
13 Contribution to TTS development:Consulting and technologies
14 Semantic solutions
15 Semantic Search: Enterprise Semantic Search solution for document system and knowledge management systems
16 Linked Data for Semantic Search: Creation-ReUse of multilingual ontologies,Linking to LOD resources,Deploying LOD
17 Linked (Open) Data for Enterprise Search
18 Semantic Search Platform
19 Customer Voice Analytics: Automatic classification of customer surveys (answers to open questions) and verbatim (customer cases or call transcriptios)
20-21 Multilingual management of verbatim coding
22 Product lines (Blogmeter, Crosslibrary)
23 Social Media Monitoring, Analytics & Management Tools per Aziende & Agenzie.
24 Blogmeter: Leader in Italia nella social media intelligence,Tecnologie d’avanguardia per la social intelligence
25 Digital Humanities e Scuola Digitale
26 Leggere i classici usando il digitale
27 I Promessi sposi e Pinocchio
28 Grazie per l’attenzione!
29 Vittorio Di Tomaso ditomaso@celi.it
2. @copyright 2014 CELI / Me-Source / Cross Library
Natural
Language
Processing
We develop software solutions based on NLP.
We are active in the Italian and International markets:
semantic search, speech technology, social media
intelligence and digital humanities.
We provide systems for intelligent management and
retrieval of un-structured information to complex
organizations (private and public sector)
2
3. @copyright 2014 CELI / Me-Source / Cross Library
4CELI’s offices
Torino
Milano
Trento
Roma 6Countries in which we operate
Italia
Belgio
Francia
Spagna
Corea
Polonia
50
People
>100
Active customers
4
Business lines
15Years of experience
NLP components
Speech technology
Social Media Intelligence
Digital Humanities
3
4. @copyright 2014 CELI / Me-Source / Cross Library
>50Published scientific papers
15Research projects
Close relationship with scientific community
6Partners in Academia
Scuola Normale Superiore
Università di Torino
Università di Pisa
Università di Trento
Fondazione Bruno Kessler
Politecnico di Milano
4
5. @copyright 2014 CELI / Me-Source / Cross Library
1999
CELI srl is
founded
1999 2005 2010
2002
Speech Technology
practice
2006
BlogMeter is
launched
2013
Launch into
Korea market
2011
Cross Library
2010
Milano, Roma,
Trento
5
8. @copyright 2014 CELI / Me-Source / Cross Library
NLP
technology
Comprehensive suite of multilingual
components and resource:
• Text processing
• Language identification
• Tokenization
• Linguistic analysis (lemmatization, POS
disambiguation, chunking) and phonetic transcription
(including intonation and prosody)
• Semantic annotations
• Named Entities and Concept extraction
• Mood detection and Sentiment analysis
• Emotion detection
8
9. @copyright 2014 CELI / Me-Source / Cross Library
Tokenization
Organization Product Date Named Entities
PN V (3_sing) N (PLU)
ADV
N (sing) P
ADV
N (sing)
ADJ
PREP
CONJ
Morphology
PN V N N PADJ PREP Disambiguation
S
ADJ N
PP
VP
V
NP
NP
PRP N
Syntactic
chunking/parsing
Semantics
PhoneticstS"elI "es "A: "el pr@v"aIdz g"Ud "en "el p"i: s@l"u:Sn=z s"Ins naInt"i:n n"aInti n"aIn
Celi provides good NLP solutions since 1999 .S.r.l
Linguistic processing and annotation
9
10. @copyright 2014 CELI / Me-Source / Cross Library
CELI’s software
solutions for multilingual
text processing and
analysis
Output: knowledge
(examples)
news
emails
verbatims
CRM tickets
social media
customer
feedback
agent notes
documents
surveys
chat
etc.
questo
documento
этий
документ
Input: multilingual
unstructured text
information
linguistic
analysis
semantic
clustering
semantic
analysis
term
/name
extraction
automatic
classification
opinion
monitoring
cross-
language
information
retrieval
names
products, people, places, etc.,
opinions
preferences of your consumers
discoveries
new unpredicted information
phonetic transcription
pronunciation representation for speech
classified texts
classification according to topics/problems:
credit card problems, unsatisfied caller,
access to an account, etc
From text to Knowledge
10
11. @copyright 2014 CELI / Me-Source / Cross Library
Meaningful intelligence from unstructured information
11
12. @copyright 2014 CELI / Me-Source / Cross Library
Comprehensive suite of
multilingual components and
resources for text processing
in Voice application (Text To
Speech)
• Grapheme to phoheme converter
designed to be used in embedded
systems
• Phonetic lexica and annotaded
corpora (both text corpora and
speech corpora)
• Coverage of 15 languages
projects
consulting
Speech
technology
12
13. @copyright 2014 CELI / Me-Source / Cross Library
TTS modules
NLP module
Consulting and technologies
text preprocessing
morphological / syntactic analysis
letter-to-sound phonetic transcription
prosody generation
Voice generation Module
acoustic database creation/annotation
unit selection algorithms
acoustic processor
Multiligual input text
Synthesized
speech
Overall linguistic feasibility study and design
Lexical resources, text corpora
Morphological / Syntactic analyzer
Phonetic transcription grammars
Prosody-annotated corpus
Voice recording assistance
Acoustic database annotation assistance
Quality assessment
Evaluation/comparison of competing products
Multilingual text preprocessing methods
Contribution to TTS development
13
15. @copyright 2014 CELI / Me-Source / Cross Library
Semantic
Search
Enterprise Semantic Search solution for
document system and knowledge management
systems
• Java platform, enterprise ready
• Full text search based on Apache Lucene
• Linguistic and semantic analysis for document
enrichment and classification
• Linguistic and semantic analysis for natural language
query understanding
• Onthologies and thesauri to improve search results,
navigation and discovery
15
16. @copyright 2014 CELI / Me-Source / Cross Library
Creation-ReUse of multilingual ontologies
• Query expansion
• Hierarchical facets
• Cross-Language Information Retrieval
Linking to LOD resources
• Content enrichment
• Discovery search
• inference
Deploying LOD
• Use of standard data models and schemas
• ETL to triple stores
• Data integration
projects
SaaS
consulting
Linked
Data for
Semantic
Search
16
17. @copyright 2014 CELI / Me-Source / Cross Library
Lorem ipsum dolor sit amet, soldatino
consectetur, sed do eiusmod tempor incididunt ut
labore et Roma magna aliqua. Ut enim ad
Luigi Einaudi, quis nostrud exercitation
ullamco laboris nisi ut aliquip ex ea commodo
consequat
http://it.dbpedia.org/resource/Luigi_Einaudihttp://purl.org/bncf/tid/17802
http://sws.geonames.org/3169071/l
i
v
e
d
I
n
Content Enrichment
Relazioni per Discovery Search
Linked (Open) Data for Enterprise Search
17
18. @copyright 2014 CELI / Me-Source / Cross Library
Service Level
Processing layer / Motore di Ricerca
Discovery
Source
Management
Onthology
management
Admin Tools
Linguistic Analysis
Lucene
Index
Config /
Monitoring
System
Expert
Ling
Resource
Authentication/
Access control
Browsing
Data
Storage
Operato
ri
Responsive Presentation
WidgetPages
API
Portali
Portals
End
users
Harvester
Adapter Adapter
Staging area
Data collection layer / Harvester
Semantic Search Platform
18
19. @copyright 2014 CELI / Me-Source / Cross Library
Customer
Voice
Analytics
Automatic classification of customer surveys
(answers to open questions) and verbatim
(customer cases or call transcriptios)
• Java platform, enterprise ready
• Available as a service or on premises
• Linguistic analysis for classification rules
• Self service
• Ready for multilingual contact centers
• Speech Analytics for quality management in contact
centers
19
20. @copyright 2014 CELI / Me-Source / Cross Library
Development of infrastructure integrated with the client’s CRM
system that manages continuous flux of multilingual information
and its automatic classification
The client receives automatically classified information for all
requested languages and has a view of its customer satisfaction in
significantly reduced time and costs of translation
Provide classified information received
in many different languages from
customer care centers located in
different countries to a marketing
department
OBJECTIVES APPROACH
Create common taxonomy for all
languages taking into account cultural
differences
Develop software and lingware for
analysis of high volume of data in
different languages
Organize a team of language experts for
development of the multilingual
resources and quality check of results
RESULTS
Multilingual management of verbatim coding
20
21. @copyright 2014 CELI / Me-Source / Cross Library
With Celi’s linguistic technologies
CRM
System
CELI
Multilingual
Classification
Service
Raw text
messages
Classification
data
Multiligual
call-centers
Notes and
tickets in
multiple
languages
Unified
Reports
Multilingual
language
technology
integrated
with
Enterprise
CRM System
CRM
System
Translators
Without Celi’s technologies
Inconsistent
Reports
Multiligual
call-centers
Multilingual management of verbatim coding
21
24. Leader in Italia nella social media intelligence
500+
Progetti realizzati
4 Miliardi
post e interazioni
social misurate l’anno
20 mila
Chiave di ricerca
configurate
7 mila
Profili aziendali social
analizzati giornalmente
80
Clienti
3
Sedi: Milano,
Roma e Torino
Tecnologie d’avanguardia per la social intelligence
Blogmeter
24