Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Iletken recommendation technologies solution

2,925 views

Published on

iletken recommendation technologies
iletken tavsiye sistemleri tanıtım

Published in: Technology, Business
  • Be the first to comment

Iletken recommendation technologies solution

  1. 1. Social Recommendation Technologies<br />
  2. 2. Recommending items of interest to users<br />based on explicit or implicit preferneces<br />Problem?<br />It is the browsing that holds the golden opportunity for a recommendation system, because the user is not focused on finding a specific thing – she is open to suggestions. <br />Alex Iskold, ReadWriteWeb 2007<br />
  3. 3. User Frustration…<br />…Lost Business Opportunity<br />
  4. 4. with<br />Increase Usage and Sales between %10-50<br />by<br />connecting<br />the right content<br />the right user<br />* iletken for Mobile Content Recommendations slide<br />
  5. 5. You Need To<br />Understand the User<br />Understand the Content<br />For Giving<br />Right Content<br />to the <br />Right User<br />
  6. 6. Content<br />Social & User Network<br />User action<br />iletken Recommender System<br />Interactions<br />Content and Context<br />Customized Solution<br />Business <br />Client<br />Analytics and Feedback<br />Real Time Recommendations<br />
  7. 7. Benefits<br />Monetize Niche Content<br />The bottom line is…<br />Generate Cross Sales<br />Increase Usability<br />Sales Increase<br />10% - 50%<br />Better Customer Service<br />Targeted Reach<br />… and more<br />
  8. 8. Awards and Global Recognition<br />3rd best recommender startup at ACM’s RecSys’08…<br />… out of 26 projects from 15 countries worldwide<br />“GeleceğıninternetindeTürkimzası.” <br />CNN Türk ’08<br />“One of 5 early recommendation technologies that could shake up their niches.”<br />ReadWriteWeb ‘08<br />iletken is a proud software partner of intel<br />iletken R&D is supported by TÜBİTAK<br />
  9. 9. Our Hybrid Technology<br />Behavior based<br />Content based<br />Social Relevancybased<br />Context based proximity graphs<br />Natural language processing<br />Collaborative filtering<br />Metadata analysis<br />MachineLearning<br />vs<br />
  10. 10. About iletken Technologies<br />
  11. 11. iletken for Media Content Recommendations<br />
  12. 12. iletken for Mobile Content Recommendations<br />Personalized targeting for…<br />Life – Ukraine results<br />… mobile game downloads and melodies<br />%331 Elevation on Niche Content<br />%411 Elevation on Popular Content<br />Overall %35-50 increase in subscription<br />
  13. 13. iletken for E-Commerce Recommendations<br />
  14. 14. Management Team<br /> Selçuk ATLI - CIO<br /><ul><li>Semantic Web and Recommender Systems LAB, TW
  15. 15. Fulbright Scholar and M.S. Information Technology @ RPI</li></ul>M. Deniz OKTAR - CEO<br /><ul><li>Founded ReklamGiy</li></ul>Barış Can DAYLIK - CTO<br /><ul><li>Natural Language Processing & MachineLearning
  16. 16. Pardus commiter</li></li></ul><li>Thanks<br />Contact info@iletken-project.com<br />Visit http://www.iletken-project.com/<br />Next: More on recommender technology<br />
  17. 17. Next: More on RecommendationTechnologies<br />1. Real WorldExample: Salesman<br />2. RecommendationMethodsDetailed<br />
  18. 18. The Salesman Analogy<br />
  19. 19. A salesmen is a Recommender<br />Recommending the right house <br />for the <br />right family<br />Difficult but why? <br /><ul><li>Needs to know about the item
  20. 20. Needs to know about the buyers
  21. 21. Needs experience </li></li></ul><li>Understand the Content - Content based filtering<br /><ul><li>My knowledge:I have a 3 room, luxury house</li></ul>Understand Users - Collaborative filtering<br /><ul><li>My Experience:If the customer lived in NYC, she will live in NYC
  22. 22. My Experience:One that bought a car is likely to buy a house
  23. 23. My Experience:Customersthatare notmarriedrents</li></li></ul><li>iletken’s award winning social approach<br />
  24. 24. İletken’s Recommendation Technology Solutions Detailed<br />Over 15 Recommendation Algorithms<br /><ul><li>Content Based
  25. 25. Collaborative Based
  26. 26. Social Based</li></ul>Developed , Tweaked & Combined<br />For each spesific business<br /><ul><li>Mobile Operator Recommendations
  27. 27. Music/Video Recommendations
  28. 28. E-Commerce Recommendations</li></li></ul><li>İletken’s Trust Networks<br />Wisdom of the Crowds<br />Circle of Trust<br />
  29. 29. Semi-ExclusiveTrust Networks<br />Let’s ask Keith about music<br />
  30. 30. Semi-ExclusiveTrust Networks<br />Trust each user for a spesific field<br />Let’s ask Keith about politics<br />He might be your expert on music but definetly not politics !<br />
  31. 31. Semi-ExclusiveTrust Networks<br />Different trust networks for different areas of interest<br />Rock and Roll<br />Politics<br />Soccer<br />
  32. 32. Two Collaborative Filtering SystemsExample<br />1. Neighboring based methods<br />2. Matrix Factorization methods<br />
  33. 33. iletken’s Semi-Exclusive Neighbor Algorithm<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />MelodyServicesProximity<br />
  34. 34. iletken’s Semi-Exclusive Neighbor Algorithm<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />+<br />Java Games Proximity<br />
  35. 35. iletken’s Matrix Factorization Methods<br />Factor 1<br />Factor 1<br />Factor 2<br />Factor 2<br />Factor 3<br />Factor 3<br />Factor 4<br />Factor 4<br /> Data driven relevancy factors<br />
  36. 36. Thanks<br />Contact info@iletken-project.com<br />Visit http://www.iletken-project.com/<br />Next: Time to contact iletken<br />

×