Making a simple question into a complicated query

1,158 views

Published on

0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,158
On SlideShare
0
From Embeds
0
Number of Embeds
4
Actions
Shares
0
Downloads
4
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

Making a simple question into a complicated query

  1. 1. Making a simple question into a complicated query Richard Boulton Lemur Consulting Ltd
  2. 2. Making a simple question into a complicated query Richard Boulton Lemur Consulting Ltd
  3. 3. Making a simple question into a complicated query Richard Boulton Lemur Consulting Ltd
  4. 4. Only 20 minutes until Lunch
  5. 5. Where shall we have Lunch?
  6. 6. “Lunch”
  7. 7. Assertion Complicated questions are easier to answer well.
  8. 8. Assertion Complicated questions are easier to answer well accurately.
  9. 9. Restaurant Pizza restaurant near Covent Garden, fairly cheap.
  10. 10. Time for a real example http://mydeco.com/ Interior decoration site
  11. 11. Users type: “Sofa” We'd prefer them to ask questions like: “Red velvet, three seater, sofa, from a supplier who can deliver to central Cambridge at a weekend”. How can we move to this kind of search?
  12. 12. Getting more from users
  13. 13. Getting more from users Suggested search completions
  14. 14. Getting more from users Facets
  15. 15. Getting more from users Facets Which facets to display? − Depends on the user. Which facet values are interesting? − A particularly fun problem for continuous numeric values, like price. How many values should we display? − Based on likelihood of any being useful?
  16. 16. Getting more from users Personal data − Using details about the user directly. e.g., Postcode − Grouping users by similarity of interests
  17. 17. Getting more from users Similarity search − “More like this” − Colour / image-based similarity
  18. 18. Behind the scenes Applying our own bias. − Perhaps we want to push some items − Perhaps we want to avoid other items − Perhaps some items go well together − Behave like a shop assistant − “Product Rank”
  19. 19. Behind the scenes Categorisation − User asks for “Sofa”. − We search for “Products categorised as one of the sofa subcategories, based on the output of a machine learning system trained with some human judgements”.
  20. 20. Behind the scenes Variety − Don't display lots of very similar items − Give the user a choice − But don't display irrelevant junk, either! − Need some way to measure variety
  21. 21. Answering complicated questions
  22. 22. Answering complicated questions Getting the best answer − Good models − Careful design − Lots of tuning
  23. 23. Answering complicated questions Getting an answer quickly − Good algorithms − Well matched data-structures − Plenty of machines − Plenty of RAM
  24. 24. Questions, and then Lunch! Richard Boulton Lemur Consulting Ltd richard@lemurconsulting.com

×