When to Consider Semantic Technology for Your Enterprise

404 views

Published on

This presentation was give by Dave Read and Michael Delaney from Blue Slate Solutions at the Semetech Technology and Business Conference in NYC on October 2nd 2013.

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
404
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

When to Consider Semantic Technology for Your Enterprise

  1. 1. © Blue Slate Solutions 2013 When to Consider Semantic Technology for Your Enterprise Michael Delaney, Senior Consulting Software Engineer David Read, CTO Semantic Technology and Business Conference New York, NY, US October 2, 2013
  2. 2. HowWhy Not
  3. 3. Who are Dave and Mike? • Architecture • Security • Innovation • Solution engineering • Integration • ETL David Read Michael Delaney
  4. 4. © Blue Slate Solutions 2013 Who is Blue Slate? 3 About  Driven by a total commitment to customers, company, colleagues and community  30+ consultants – operations, strategy, technology and industry experts  Founded in 2000, headquartered in Albany, NY Clients  Industry Leaders seeking to drive value for shareholders and end customers  Include ten Blues organizations and four commercial payers  Innovators looking to grow beyond their core markets
  5. 5. © Blue Slate Solutions 2013 What? 4 Let’s give semantic technology some context
  6. 6. © Blue Slate Solutions 2013 What is Semantic Technology? Semantics ≡ meaning Semantic Technology ≡ machine-readable meaning 5
  7. 7. © Blue Slate Solutions 2013 What Makes Up Semantic Technology (in this talk)? • Standards → RDF/RDFS/OWL/SPARQL • Definitions → Ontologies • Storage → Triple Stores • Inferencing → Reasoners • Data Access → SPARQL • APIs → Jena, Sesame 6
  8. 8. © Blue Slate Solutions 2013 Semantic Technology is a Team Player in an Architecture • Integrates • Federates • Adapts • Extends 7
  9. 9. © Blue Slate Solutions 2013 What Is Different About Semantic Technology? • Structure is (mostly) logical not physical – Triple – Directed Graph • Federation is assumed – SPARQL • Web is the natural platform – URI – HTTP 8 Subject Predicate Object
  10. 10. © Blue Slate Solutions 2013 Directed Graph Example 9 David bioFatherOf Sarah Lisa bioMotherOf fullSiblingOf Michael SarahB friendOf friendOf Carl spouseOf Blue Slate Solutions employeeOf favoriteSport Bowling bioFatherOf fullSiblingOf
  11. 11. © Blue Slate Solutions 2013 The Physical Structure is Flexible by Design • Triples are an extreme form of normalization • Any data can be related without the need for foreign keys • Relationships can be added or removed as they are found, explored, accepted or discredited 10 Lisa bioMotherOf Michael friendOf Carl favoriteSport Bowling
  12. 12. © Blue Slate Solutions 2013 The Logical Structure is Built to Relate and Define • Directed graph relates data • RDF, RDFS and OWL define data • Reasoners build models on the fly • Relationships are flexible: – Groups can use different relationship rules 11
  13. 13. © Blue Slate Solutions 2013 What’s New? Let’s compare semantic and other technologies 12
  14. 14. © Blue Slate Solutions 2013 Storage 13 Relational Efficient Use of Space NoSQL Document Stores (Key-Value Pairs) Semantic Single, standardized schema (the triple)
  15. 15. © Blue Slate Solutions 2013 Relationships 14 Relational Relationships convey meaning through table names or data values NoSQL No Semantic Relationships convey meaning in their definition
  16. 16. © Blue Slate Solutions 2013 Query Language 15 Relational Standardized Language (ISO: SQL) NoSQL No Language Standardization Semantic Standardized Language (SPARQL)
  17. 17. © Blue Slate Solutions 2013 Schema Maintenance 16 Relational Rigid Schema NoSQL Data is unstructured Semantic Relationships convey meaning in their definition
  18. 18. © Blue Slate Solutions 2013 Maturity 17 Relational Well established (30+ years old) NoSQL Coming Along (10+ years old) Semantic Relatively Young (5+ years old)
  19. 19. © Blue Slate Solutions 2013 Indexing 18 Relational Indexing can be complicated, but is vital to scalability NoSQL Indexing can be complicated, but is vital to scalability Semantic Indexing can be automatic
  20. 20. © Blue Slate Solutions 2013 Optional Relationships 19 Relational Optional relationships can be difficult (NULL’s or Dummy Values) NoSQL None Semantic All relationships are optional by default, but can be enforced if needed
  21. 21. © Blue Slate Solutions 2013 Ecosystem 20 Relational Very rich development ecosystem, solid tooling NoSQL Rich development ecosystem, limited tooling Semantic Immature ecosystem, nascent tools
  22. 22. © Blue Slate Solutions 2013 Ontology OWL Semantic Technology: Cool Solution Looking for a Problem? 21 Triple Graph SPARQL Reasoner Triple Store RDFTurtleObject Predicate Subject
  23. 23. © Blue Slate Solutions 2013 Where? 22 Is this technology really being used?
  24. 24. © Blue Slate Solutions 2013 Where is Semantic Technology Today? • Maturing standards and practices • Expanding ecosystem • Big and small players • Diverse offerings – Products and services • Real-world usage 23
  25. 25. © Blue Slate Solutions 2013 Consider Relational Database Adoption 24 1 2 3 4 5 7 1. Edgar Codd begins relational data research 2. Codd publishes, “Relational Model of Data for Large Shared Data Banks” 3. Oracle v2 released (marketing gimmick, no v1) 4. DB2 announced 5. Sybase released 6. DB2 released 7. Sybase forked to MS SQL Server 6
  26. 26. © Blue Slate Solutions 2013 Data and Analytic Diversity Continue to Expand 25 • Evolving Interactions • Data Diversity
  27. 27. © Blue Slate Solutions 2013 Where has it been deployed successfully? 26 • BBC – World Cup – Olympics • data.gov • Facebook’s Graph Search • Best Buy Product Catalog • Cleveland Clinic • Amazon • …
  28. 28. © Blue Slate Solutions 2013 Why? Why do enterprises use semantic technology? 27
  29. 29. © Blue Slate Solutions 2013 How Does Semantic Technology Benefit an Enterprise? Integrates diverse data sources 28
  30. 30. © Blue Slate Solutions 2013 How Does Semantic Technology Benefit an Enterprise? Provides consistent meaning across systems 29
  31. 31. © Blue Slate Solutions 2013 How Does Semantic Technology Benefit an Enterprise? Federates and integrates in real-time 30
  32. 32. © Blue Slate Solutions 2013 How Does Semantic Technology Benefit an Enterprise? Supports multiple logical models without ETL and warehouses 31
  33. 33. © Blue Slate Solutions 2013 How Does Semantic Technology Benefit an Enterprise? Extends data and relationships without refactoring databases 32
  34. 34. © Blue Slate Solutions 2013 How Does Semantic Technology Benefit an Enterprise? Augments Analytics 33
  35. 35. © Blue Slate Solutions 2013 Volume, Velocity, Variability, Variety • Triple stores do well with volume and velocity – Commercial products scale to billions of triples • Directed graphs simplify variability – Nodes and vertices can come and go • URIs support variety – Addressable ≡ accessible 34
  36. 36. © Blue Slate Solutions 201335 When? Recognizing good opportunities for introducing semantic technology
  37. 37. © Blue Slate Solutions 2013 When should you consider Semantic Technology? Heterogeneous Data 36
  38. 38. © Blue Slate Solutions 2013 When should you consider Semantic Technology? 37 Evolving Data Structures
  39. 39. © Blue Slate Solutions 2013 When should you consider Semantic Technology? 38 Rule-based data interactions
  40. 40. © Blue Slate Solutions 2013 When should you consider Semantic Technology? 39 Language Flexibility
  41. 41. © Blue Slate Solutions 2013 When should you consider Semantic Technology? 40 Vendor Independence
  42. 42. © Blue Slate Solutions 2013 When should you consider Semantic Technology? 41 Augmenting Current Systems
  43. 43. © Blue Slate Solutions 2013 When Not? 42 Avoid the round hole, square peg syndrome
  44. 44. © Blue Slate Solutions 2013 When Won’t Introducing Semantics Make a Splash? • Common and well-defined systems – Claims Processing – Order Entry – ERP – Payroll – CRM 43
  45. 45. © Blue Slate Solutions 2013 Semantic Technology is Not “The” Answer Not a dessert topping, engine lubricant, elixir-of-life, hand soap and cloud solution all-in-one! 44
  46. 46. © Blue Slate Solutions 2013 When is Semantic Technology a bad fit? Transactional, high volume 45
  47. 47. © Blue Slate Solutions 2013 When is Semantic Technology a bad fit? 46 Homogeneous or contained data
  48. 48. © Blue Slate Solutions 2013 When is Semantic Technology a bad fit? Problems that have already been solved/abstracted 47
  49. 49. © Blue Slate Solutions 201348 When is Semantic Technology a bad fit? Copying (large) data stores
  50. 50. © Blue Slate Solutions 2013 Case Study 49 We’re giving you the story first hand
  51. 51. © Blue Slate Solutions 2013 Case Study: Client • Medicare Administrative Contractor • Struggling with key metrics • Competitive landscape (15  10  ~7) 50
  52. 52. © Blue Slate Solutions 2013 Case Study: Challenge • Poor medical review focus – Only denying ~30% of reviewed claims • Slow reaction to review outcomes • High repeat defects from providers • Struggling to implement meaningful analytics 51
  53. 53. © Blue Slate Solutions 2013 Case Study: Root Causes • Manual “copying” between systems • Lack of agility around review checklists • Review details not captured • Providers mystified regarding denial reasons • Multiple systems with parts of the story 52
  54. 54. © Blue Slate Solutions 2013 Case Study: Semantic Focus • Agile checklist support • Flexible data classifications for denials • Immediate aggregation and feedback of reviews • Legacy integration 53 • Detailed association of denials and provider education • Integration of claim data and reviews for analytics
  55. 55. © Blue Slate Solutions 2013 • Details support targeted data mining • Ontology easily updated to address identified risks • Semantic reasoner leverages rules discovered through analytics • Flexible classifications provide agility dealing with changing healthcare payer issues (errors, fraud, abuse, waste…) Case Study: Outcomes 54
  56. 56. © Blue Slate Solutions 2013 Case Study: Results (Just Published) 55 Activity Established Goal Current Results Probe (Hypothesis) Research Duration 250% Improvement 370% Improvement Last Claim Reviewed to Date of Notification 240% Improvement 2070% Improvement (months to days) Probe Edits Above 35% Charge Denial Rate 62% 100% After 9 months of production experience…
  57. 57. © Blue Slate Solutions 2013 How? 56 Make it so!
  58. 58. © Blue Slate Solutions 2013 How do I know when to take the leap? 57
  59. 59. © Blue Slate Solutions 2013 How do I start? Assign an Owner 58 Pick a Project Iterate over Technology
  60. 60. © Blue Slate Solutions 2013 Assign an Owner 59 • Point person • Responsible and accountable • “Gets” the big picture value proposition • Educator • Evangelist
  61. 61. © Blue Slate Solutions 2013 Pick a project 60 • Aligned with semantic technology value proposition • Creative team • Education investment • Agile process • Visible • Schedule flexibility (phases)
  62. 62. © Blue Slate Solutions 2013 Iterate over Technology 61 • Learn • Try • Select • Full speed!
  63. 63. © Blue Slate Solutions 2013 Questions + Contact Info 62 Thank you for attending our session. Feel free to contact us if you have questions: Michael.Delaney@blueslate.net David.Read@blueslate.net www.blueslate.net

×