Country Clustering based onSearch-Query Pattern CorrelationEdgar Anzaldúa MorenoRiaz Esmailzadeh
About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing Department
About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing Department...
About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing Department...
About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing Department...
Tracking People• Watch People Closely• Geographically Distant Countries• Keywords & Search Engines
The World Today
The World Today
The World Today?
Long Tails
Long Tails
Long Tails
Long TailsCountry	  ACountry	  BCountry	  CCountry	  D0% 25% 75% 100%
Long TailsCountry	  ACountry	  BCountry	  CCountry	  D0% 25% 75% 100%Country	  ECountry	  FCountry	  GCountry	  H
Pearson
Pearson
Correlation
Correlation
Correlation
Correlation
Correlation
Similarity
Similarity
Similarity
Similarity
Country Graph
Country Graph
Country Graph
Country Graph
Louvain
Results
Results vs. GDP
Future Work• Search-Query Pattern Analysis• Extract User Interests based on Search-Queriesand Countries• Predict User Beha...
Conclusions• Cultural Stereotypes don’t match User Interests• Data is not Information• There is still a lot of work to do
Questions?http://eam.mx/getthepaperGet My Card
Upcoming SlideShare
Loading in …5
×

2 | Page Country Clustering Based on Search-Query Pattern Correlation

267 views

Published on

This paper introduces a technique to cluster regions or countries for marketing purposes. It uses information gathered through aggregated website search-query patterns, available from web logs or analytics software. Clustering done using this technique helps grouping countries that similarly respond to a particular marketing campaigns potentially reducing marketing costs. The method collects aggregated geographical and search-pattern information from keywords and the visit volume of each keyword. It uses abstract keyword logs in web analytics software to identify and understand target market segments more effectively. It is argued that actual search logs could be better indication of market behavior than language or cultural boundaries. By analyzing web logs with this technique, marketers will be able to plan marketing campaigns strategically based on the empirical data on website visitors. This data is derived from huge long-tailed keyword search sources, and ultimately helps understand and adapt better to customer’s interests. The paper starts with an introduction on the underlying problem, the analysis method, and its four basic steps: preparing the data, creating a correlation matrix, identifying criteria and region clustering using the Louvain method of community structure over random networks. The results for a specific dataset are then presented and conclude with a list of future work steps. Results in this paper show that country-based keyword analysis can yield search-query patterns that unveil connections between culturally dissimilar regions.

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

  • Be the first to like this

No Downloads
Views
Total views
267
On SlideShare
0
From Embeds
0
Number of Embeds
12
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

2 | Page Country Clustering Based on Search-Query Pattern Correlation

  1. 1. Country Clustering based onSearch-Query Pattern CorrelationEdgar Anzaldúa MorenoRiaz Esmailzadeh
  2. 2. About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing Department
  3. 3. About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing DepartmentEnrolments
  4. 4. About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing DepartmentEnrolments Visits
  5. 5. About MyselfMSc Information TechnologyCarnegie Mellon UniversityWeb Persuasion ArchitectInternational Marketing DepartmentEnrolments Visits Facebook  Fans
  6. 6. Tracking People• Watch People Closely• Geographically Distant Countries• Keywords & Search Engines
  7. 7. The World Today
  8. 8. The World Today
  9. 9. The World Today?
  10. 10. Long Tails
  11. 11. Long Tails
  12. 12. Long Tails
  13. 13. Long TailsCountry  ACountry  BCountry  CCountry  D0% 25% 75% 100%
  14. 14. Long TailsCountry  ACountry  BCountry  CCountry  D0% 25% 75% 100%Country  ECountry  FCountry  GCountry  H
  15. 15. Pearson
  16. 16. Pearson
  17. 17. Correlation
  18. 18. Correlation
  19. 19. Correlation
  20. 20. Correlation
  21. 21. Correlation
  22. 22. Similarity
  23. 23. Similarity
  24. 24. Similarity
  25. 25. Similarity
  26. 26. Country Graph
  27. 27. Country Graph
  28. 28. Country Graph
  29. 29. Country Graph
  30. 30. Louvain
  31. 31. Results
  32. 32. Results vs. GDP
  33. 33. Future Work• Search-Query Pattern Analysis• Extract User Interests based on Search-Queriesand Countries• Predict User Behavior• Semantic Translation• Typo Support
  34. 34. Conclusions• Cultural Stereotypes don’t match User Interests• Data is not Information• There is still a lot of work to do
  35. 35. Questions?http://eam.mx/getthepaperGet My Card

×