ICIC 2013 New Product Introductions CEPT
2 likes1,141 views
CEPT Systems is developing a new natural language processing technology called Semantic Fingerprinting that can dramatically improve how businesses process large amounts of text-based data. Their technology, called the CEPT Retina, converts words and documents into semantic fingerprints that capture relationships between meanings. These fingerprints allow for direct comparison of word and document similarities. CEPT offers their technology as a cloud-based API that is simple for developers to use and integrate into various applications. Their technology has 12 application-specific APIs and is aimed to help businesses with tasks like search, classification, discovery, and analytics using semantic analysis of text. An example success story is an online language learning company that is using the CEPT API to lower costs, improve learner motivation, generate
ICIC 2013 New Product Introductions CEPT
- 1. CEPT SYSTEMS @ ICIC 2013
- 2. CEPT Systems Inc. Who we are?
- 3. CEPT Systems Inc. based in Vienna, Austria Cortical Engine for Processing Text New technology to dramatically improve natural language processing
- 4. The impediment What is the problem?
- 5. Businesses are threatened by the BIG DATA explosion A substantial part of this data consists of text Social Nets Blogs Documents Numbers Webpages Email News
- 6. Semantic processing of natural language becomes essential for businesses... (1) to locate documents (2) to find web content (3) to match the skills of people (4) to identify and select products (5) to monitor competitors (6) to file business information (7) to drive product innovation (8) to track customer satisfaction (9) to avoid duplication of work (10) to advertise on the Internet (11) to mine for evidences (12) to improve security
- 7. The Downsides of the state-of-the-art High quality semantic systems are: •Hard to build (sometimes impossible) •Inaccurate and fragile (in real-world use) •Expensive to buy (licenses and/or services) •Tricky to integrate (setup, tuning, training ...) •Laborious to run (metadata management) •Time-consuming to maintain (dict., ontologies)
- 9. banana apple computer The CEPT Retina converts any word into its Semantic Fingerprint
- 10. cat cat + dog dog The fingerprints allow direct semantic comparison of the meanings of any two words
- 11. cat cat + dog Home and family aspects dog Biology aspects Leisure and free time aspects Cat specific Dog specific Social aspects The CEPT-Retina shows thousands of semantic relations for any two words
- 12. word fingerprints document fingerprint Word Fingerprints are aggregated into Document Fingerprints
- 13. Example documents provided by user for a specific field of interest document fingerprints profile fingerprint Document Fingerprints are aggregated into User-Profile Fingerprints
- 14. Similar Documents Result-Set query fingerprint Similarity Engine Profile FP Profile Ranking Fingerprints of indexed document collection The high performance similarity algorithm provides the Semantic Search Engine functionality
- 15. Jaguar - Porsche SUB = Tiger RETURN Expression Engine The Expression Engine permits arithmetic computation with fingerprints
- 16. CEPT technology is offered as cloud based SaaS ✓It is simple to use, efficient and cost-effective ✓It can be easily embedded in diverse applications ✓It is licensed to developers as a web-service API ✓It is offered as transaction-based subscription-model
- 17. Strategy to ensure developer adoption ✓ No Natural Language Processing experience needed by developers ✓ Existing domain developers can immediately start using it ✓ Business people can be part of the development team ✓ CEPT-API is offered as 12 different flavors, each one optimized for a specific business case
- 18. Overview 12 Disciplines of Information Retrieval 1. 2. 3. 4. 5. Enterprise Search - Document Management Website Search - Marketing, Sales, Support Profile Matching - Recruiting, Marketplaces, Dating Product Search - E-Commerce, Retail, Wholesale Alerting Systems - Competitive Intelligence 6. 7. 8. 9. Content Classification - News Filtering, Document Classifier Information Discovery - Innovation Support Opinion Mining - Social Network / Peer groups Analysis Office Automation - Intelligent Office Productivity System 10. Keyword Generation - Search Engine Optimization 11. Forensic Text Analytics - Evidence Mining, Plagiarism Detection 12. Intelligence & Security - Real Time Message Monitoring
- 19. CEPT Service Technology Stack Demos cortical.io Demos Example Code Example Code and and Widgets Widgets Tutorials Discipline 2 Website Search API CEPT API Retina En De Fr Sp Chem Med Bio Pharm Monitoring Expression Engine Accounting Demos Example Code Example Code and and Widgets Widgets ... Tutorials Discipline 1 Enterprise Search API Demos Tutorials Tutorials Discipline 11 Forensic Analytics API Discipline 12 Intelligence Security API Similarity Engine Key Management Identity Management Service Management Layer (3Scale) Elastic Beanstalk Dynamo DB Simple Workflow Service Cloud Infrastructure (Amazon) Elastic Block Store
- 20. A first Success Story Online Language Learning
- 21. Online Sprachlernen ... mehr als 1 Million Benutzer
- 22. ✓ ✓ ✓ ✓ Lowering Cost: The most expensive part of producing language learning materials are the pedagogic editors. Phase-6 can lower these costs by maintaining a repository of 1Billion high quality example sentences, that are quality checked using the CEPT-API. Improve Learner Motivation: The learner can choose a topic of interest and all the examples, exercises and tests are generated within this topic using the CEPT-Retina. Individualized Tests: Traditional e-learning systems just discriminate between „right“ and „wrong“. Using the CEPT-API allows Phase-6 to give more specific answers like „nearly right“ or „you are on your way“. Intelligent Tutoring: Using the CEPT-API the Phase-6 system can give intelligent hints, depending on the learners chosen topic, his skill level and the kind of error he made before.