Semantic content enrichment and techniques for extraction of
- Slides: 55
Semantic content enrichment and techniques for extraction of relevant relation (part 2) Presentation of several uses-cases, their link to the Big Data and NLP community Charles Huot & Stefan Geissler Institute for Bulgarian Language, Bulgarian Academy of Sciences Sophia March 5 th, 2014 Copyright © 2014 TEMIS - All Rights Reserved - Slide 1
Copyright © 2014 TEMIS - All Rights Reserved - Slide 2 It’s all about languages, isn’t it?
Mc. Kinsey Global Institute: « Big data: The next Frontier for innovation, competition, and productivity Copyright © 2014 TEMIS - All Rights Reserved - Slide 3
« Nation’s most pressing challenges » Copyright © 2014 TEMIS - All Rights Reserved - Slide 4
More Data, much more Copyright © 2014 TEMIS - All Rights Reserved - Slide 5
Copyright © 2014 TEMIS - All Rights Reserved - Slide 6
Connected Cows Copyright © 2014 TEMIS - All Rights Reserved - Slide 7
Sensors on airplane engines Copyright © 2014 TEMIS - All Rights Reserved - Slide 8
Sensors to monitor the trafic and more in town Copyright © 2014 TEMIS - All Rights Reserved - Slide 9
Collecting data from the Universe Copyright © 2014 TEMIS - All Rights Reserved - Slide 10
Génome Decoded Genome Copyright © 2014 TEMIS - All Rights Reserved - Slide 11
Crowd Sourcing Copyright © 2014 TEMIS - All Rights Reserved - Slide 12
Play, Buy, communicate s and leave numerical data in all places Copyright © 2014 TEMIS - All Rights Reserved - Slide 13
Talk about ourselves and others Copyright © 2014 TEMIS - All Rights Reserved - Slide 14
Size Matter, but it’s not the only criteria ! 1 Tone of electronic equipment The largest Hard Drive Of the IBM latest generation computer, the “ 305 RAMAC” in 1956, with 5 Mo capacity Copyright © 2014 TEMIS - All Rights Reserved - Slide 15
More applications to manage Watson Number & complexity of technologies/data sources Predictions Gov’t Intelligence Investment Trend Decision support Apps Detection Fraud Detection Voice of Customer Sentiment extraction Relationship Detection Climate Modeling And Prediction Brand management e. Discovery Reputation management Find drug interactions Find influencers Alerting Pattern Log Analysis Detection Ad targeting Smarter Planet Churn detection e. Commerce Time Frame Historic Source: IDC, Big Data and the information advantage Sue Feldman Copyright © 2014 TEMIS - All Rights Reserved - Slide 16 Future(Predict)
Solution d’analyse des verbatim clients 3. Restitution 2. 1. Surveys Multichannels feedback collection Experts 2. Verbatim Analysis Verbatim Customers IT Users Copyright © 2014 TEMIS - All Rights Reserved - Slide 17
La plateforme Luxid Annotation Factory, chaîne de traitement en langage naturel. 2. Analyse des verbatim Verbatim SC A Skill Cartridges® Anonymisation SC SC SC Skill Cartridges® Métier Bancaire Skill Cartridges® Analyse Opinions Skill Cartridges® Propos Diffamatoires B O Content Enrichment Studio Copyright © 2014 TEMIS - All Rights Reserved - Slide 18 P Experts
Meta data Structuration Metadata attached to verbatim are 4 types : 1. données qualifiant la nature de la source, le mode de recueil, la date, le type d'enquête, le marché, libellé question 2. données recueillies lors de l'enquête : note et verbatim 3. données d'identification du client s'il y a levé l'anonymat 4. données de localisation de l'agence du client Copyright © 2014 TEMIS - All Rights Reserved - Slide 19
1. Anonymisation of content SC A Skill Cartridges® Anonymisation Copyright © 2014 TEMIS - All Rights Reserved - Slide 20
2. Banking Business Model SC * B Skill Cartridges® For Banking Copyright © 2014 TEMIS - All Rights Reserved - Slide 21
3. Opinions Mining Sentiments detection SC O Skill Cartridges® Analyse Opinions Copyright © 2014 TEMIS - All Rights Reserved - Slide 22
Copyright © 2014 TEMIS - All Rights Reserved - Slide 23
Copyright © 2014 TEMIS - All Rights Reserved - Slide 24
Copyright © 2014 TEMIS - All Rights Reserved - Slide 25
Copyright © 2014 TEMIS - All Rights Reserved - Slide 26
Copyright © 2014 TEMIS - All Rights Reserved - Slide 27
Copyright © 2014 TEMIS - All Rights Reserved - Slide 28
Copyright © 2014 TEMIS - All Rights Reserved - Slide 29
Copyright © 2014 TEMIS - All Rights Reserved - Slide 30
Copyright © 2014 TEMIS - All Rights Reserved - Slide 31
TV & Twitter Copyright © 2014 TEMIS - All Rights Reserved - Slide 32
Luxid Navigator Biopharma Edition: Big Data & Open Data Copyright © 2014 TEMIS - All Rights Reserved - Slide 33
Copyright © 2014 TEMIS - All Rights Reserved - Slide 38
Copyright © 2014 TEMIS - All Rights Reserved - Slide 39
Copyright © 2014 TEMIS - All Rights Reserved - Slide 40
Copyright © 2014 TEMIS - All Rights Reserved - Slide 41
Copyright © 2014 TEMIS - All Rights Reserved - Slide 42
The success factor? Emergence of eco -system to ensure data interoperability Copyright © 2014 TEMIS - All Rights Reserved - Slide 43
The first collaborative semantics platform Our goal : to create synergies in our ecosystem Platform Partners Service Partners Academic Partners Clients Copyright © 2014 TEMIS - All Rights Reserved - Slide 44
Copyright © 2014 TEMIS - All Rights Reserved - Slide 45
Copyright © 2014 TEMIS - All Rights Reserved - Slide 46
Luxid® Community & LT-Innovate What can LT-Innovate do to help? v Luxid® is an open platform v Luxid® Community is an open initiative • Any one can contribute v Platform partner program in place • Free access to the tools (annotation platform & studio) • Ability to promote services • Ability to sell products (CC-By framework) v Contribution from LT-Innovate? • Join the community to use and evaluate existing resources • Enrich resource portfolio - post industry oriented thesaurus • Contribute to Semantics - build and share Skill Cartridges™ Copyright © 2014 TEMIS - All Rights Reserved - Slide 47
Summary Open the door to the Digital Single Market v NLP is the key • Provides a language independent knowledge representation • Skill Cartridges™ as the foundations v Unlock seamless & multi-lingual search & navigation • Local data (customer interactions, social media, …) • Global data (technical & scientific knowledge, …) v Unlock data interoperability • Collaboration between industry & data owner is required • Luxid® Community as an enabler/accelerator Copyright © 2014 TEMIS - All Rights Reserved - Slide 48
Copyright © 2014 TEMIS - All Rights Reserved - Slide 49
Copyright © 2014 TEMIS - All Rights Reserved - Slide 50
Copyright © 2014 TEMIS - All Rights Reserved - Slide 51
v Related Documents Popup • Links the reader to additional content based on the current document v « Slice and Dice » approach • Copyright © 2014 TEMIS - All Rights Reserved - Slide 52 Chapters from separate books are available on an individual basis
Commercial Rents (size, location, facilities, time, region, . . . ) Alimony/Child support (income, marriage, children, region, . . . ) Labor Law (age, company size, location, industry, region, …) Damages (the type of loss, circumstances, region, . . . ) Copyright © 2014 TEMIS - All Rights Reserved - Slide 53
Querying the Knowledge Base topic Labor Law subtopic Accidents in the workplace Type of disability Date of decision Court Age Gender Damages paid Profession Type of accident Cause Severity of the disability Part of body Copyright © 2014 TEMIS - All Rights Reserved - Slide 54 Each extracted concept has become a search criteria
Geographical Query Innovative geographical outline-based query mechanism v Goes beyond conventional search v Makes use of address extractions to focus on geographical area of interest Copyright © 2014 TEMIS - All Rights Reserved - Slide 55
Geographical Query - results Innovative geolocalized results v Goes beyond conventional and flat search results list v Makes use of address extractions to locate decisions on the map Copyright © 2014 TEMIS - All Rights Reserved - Slide 56
Navigating Query Results Decisions matching the search criteria Facets based on the same concepts enable efficient drill-down within the document set Copyright © 2014 TEMIS - All Rights Reserved - Slide 57
Navigating within 1 query result Enhanced experience with accelerated navigation v Semantic shortcuts accelerate navigation v Key concepts highlighting Copyright © 2014 TEMIS - All Rights Reserved - Slide 58
Copyright © 2014 TEMIS - All Rights Reserved - Slide 59
