SlideShare a Scribd company logo
5 useful things to do
with click data
Dr. Gavin Ruddy
gav@pontneo.com
12 Oct 2015
1: improve relevance
match with
high click rate
match with
low click rate
boost useful itemsquery
lots of items
clicking provides direct user feedback
2: broaden responses
query
intersecting
clickstream
data
extended return
click traffic identifies important connections
3: refine point of view
query focussed return
e.g. filter
by user
type
different groups can see things differently
4: knowledge sharing
e.g. things I visit things I might like
click data reflects collective understanding
5: implicit adaptation
query dynamic return
new
clicking
change
natural change via an implicit feedback loop
… overall …
lots of items
click data enables
a self-organising
“Knowledge Map”,
built and applied
from the bottom up
… so what …
• continual intelligent filtering
• flexible, scalable methods that run at query time
• does not require complex models, data or techniques
• significant, low risk improvements
• useful in many situations
• effectively controlled by how you choose to integrate
click data over time
• helps deal with many of the usual problems
associated with collaborative methods
… but it’s complicated …
• no it isn’t
• click signal can be noisy, but there’s lots of it,
it’s all real and it’s highly dynamic
• just record appropriate clicks in a useful form for
integration into your queries (e.g. fromID, toID
pairs with a timestamp)
• for more detail on how to go about applying
these methods, see “Click-Through Filter” e.g.
http://www.slideshare.net/pontneo/better-search-implementation-
of-click-through-filter-as-a-query-parser-plugin-for-apache-solr-
lucene
… but it’s complicated …
• no it isn’t
• click signal can be noisy, but there’s lots of it,
it’s all real and it’s highly dynamic
• just record appropriate clicks in a useful form for
integration into your queries (e.g. fromID, toID
pairs with a timestamp)
• for more detail on how to go about applying
these methods, see “Click-Through Filter” e.g.
http://www.slideshare.net/pontneo/better-search-implementation-
of-click-through-filter-as-a-query-parser-plugin-for-apache-solr-
lucene

More Related Content

PDF
Certificate
PDF
How To Drive Data Driven Change In A Legacy Organization
PPT
Crisis comms in the age of social media
PDF
Gabeesh Mishra - MIT Big Data Certification
PPT
Web evaluation scorecard
PDF
certificate
PPTX
Project Management: Understanding Organizational Requirements
Certificate
How To Drive Data Driven Change In A Legacy Organization
Crisis comms in the age of social media
Gabeesh Mishra - MIT Big Data Certification
Web evaluation scorecard
certificate
Project Management: Understanding Organizational Requirements

What's hot (14)

PDF
2018
PPTX
A predictive analytics primer
PDF
certificate
PPTX
Don’t Create In A Vacuum
PDF
certificado digital
PPTX
9 Steps to Advocacy Evaluation
PPTX
e-Assessment Discussing the future
PPTX
Setting up for Success: School Technology
DOCX
Real Time Evaluation Handout
PPTX
Simplify your data analytics strategy
PDF
Google: The future of apps is web
PDF
MIS5101 Week 13 Security Privacy Data Mining
PDF
MIS5001 Week 13 Technology Dependence
PDF
Why Your Data and Analytics Should Live in the Cloud
2018
A predictive analytics primer
certificate
Don’t Create In A Vacuum
certificado digital
9 Steps to Advocacy Evaluation
e-Assessment Discussing the future
Setting up for Success: School Technology
Real Time Evaluation Handout
Simplify your data analytics strategy
Google: The future of apps is web
MIS5101 Week 13 Security Privacy Data Mining
MIS5001 Week 13 Technology Dependence
Why Your Data and Analytics Should Live in the Cloud
Ad

Viewers also liked (7)

PPT
Quick introduction to the click-through filter
PPT
Better Search: Click-through filter plugin – a flexible tool for improving se...
PDF
Implementing Click-through Relevance Ranking in Solr and LucidWorks Enterprise
PPT
Amazon Item-to-Item Recommendations
PDF
Kdd 2014 Tutorial - the recommender problem revisited
PPT
Recommendation system
PDF
Recommender Systems (Machine Learning Summer School 2014 @ CMU)
Quick introduction to the click-through filter
Better Search: Click-through filter plugin – a flexible tool for improving se...
Implementing Click-through Relevance Ranking in Solr and LucidWorks Enterprise
Amazon Item-to-Item Recommendations
Kdd 2014 Tutorial - the recommender problem revisited
Recommendation system
Recommender Systems (Machine Learning Summer School 2014 @ CMU)
Ad

Similar to 5 useful things to do with click data (20)

PPTX
Digital Learning Sample Metrics
PPTX
Learning Analytics Primer: Getting Started with Learning and Performance Anal...
PDF
Data Analytics Action Figures
PDF
Data Collection Tool Used For Information About Individuals
PDF
Data Product Teams Ecosystems
DOC
How to design questionnaire
DOCX
PepsiCo Decision Support System
PDF
How to unlock new data-driven potential for your organization
PPT
Lean information
PPTX
Transforming Healthcare: Build vs Buy
 
PPTX
01 deloitte predictive analytics analytics summit-09-30-14_092514
PPTX
Feeding back diagnostic information by jeff (copy)
PPTX
Rational decision making and long term decisions
PDF
Growing a Culture of Data-Driven Continuous Improvement
PPTX
UKSG Conference 2017 Breakout - User Engagement Analytics: measuring and driv...
PPTX
Module 3 - Improving Current Business with External Data- Online
PPTX
Assessing and Advancing Community Readiness for Multi-Sector Data Sharing
PPTX
Learning Data Analytics
PPTX
UX Brighton - Analytics Best Practice
PPTX
RWJF Advancing Social Media Metrics Meeting
Digital Learning Sample Metrics
Learning Analytics Primer: Getting Started with Learning and Performance Anal...
Data Analytics Action Figures
Data Collection Tool Used For Information About Individuals
Data Product Teams Ecosystems
How to design questionnaire
PepsiCo Decision Support System
How to unlock new data-driven potential for your organization
Lean information
Transforming Healthcare: Build vs Buy
 
01 deloitte predictive analytics analytics summit-09-30-14_092514
Feeding back diagnostic information by jeff (copy)
Rational decision making and long term decisions
Growing a Culture of Data-Driven Continuous Improvement
UKSG Conference 2017 Breakout - User Engagement Analytics: measuring and driv...
Module 3 - Improving Current Business with External Data- Online
Assessing and Advancing Community Readiness for Multi-Sector Data Sharing
Learning Data Analytics
UX Brighton - Analytics Best Practice
RWJF Advancing Social Media Metrics Meeting

Recently uploaded (20)

PDF
REAL ILLUMINATI AGENT IN KAMPALA UGANDA CALL ON+256765750853/0705037305
PPTX
sac 451hinhgsgshssjsjsjheegdggeegegdggddgeg.pptx
PDF
168300704-gasification-ppt.pdfhghhhsjsjhsuxush
PPTX
Managing Community Partner Relationships
PPT
Predictive modeling basics in data cleaning process
PPTX
STERILIZATION AND DISINFECTION-1.ppthhhbx
PDF
OneRead_20250728_1808.pdfhdhddhshahwhwwjjaaja
PPTX
Qualitative Qantitative and Mixed Methods.pptx
PDF
Microsoft Core Cloud Services powerpoint
PDF
Global Data and Analytics Market Outlook Report
PPTX
Copy of 16 Timeline & Flowchart Templates – HubSpot.pptx
PPTX
SAP 2 completion done . PRESENTATION.pptx
PDF
Optimise Shopper Experiences with a Strong Data Estate.pdf
PDF
Capcut Pro Crack For PC Latest Version {Fully Unlocked 2025}
PPTX
(Ali Hamza) Roll No: (F24-BSCS-1103).pptx
DOCX
Factor Analysis Word Document Presentation
PPTX
QUANTUM_COMPUTING_AND_ITS_POTENTIAL_APPLICATIONS[2].pptx
PPTX
Database Infoormation System (DBIS).pptx
PPTX
Introduction to Inferential Statistics.pptx
PPTX
Leprosy and NLEP programme community medicine
REAL ILLUMINATI AGENT IN KAMPALA UGANDA CALL ON+256765750853/0705037305
sac 451hinhgsgshssjsjsjheegdggeegegdggddgeg.pptx
168300704-gasification-ppt.pdfhghhhsjsjhsuxush
Managing Community Partner Relationships
Predictive modeling basics in data cleaning process
STERILIZATION AND DISINFECTION-1.ppthhhbx
OneRead_20250728_1808.pdfhdhddhshahwhwwjjaaja
Qualitative Qantitative and Mixed Methods.pptx
Microsoft Core Cloud Services powerpoint
Global Data and Analytics Market Outlook Report
Copy of 16 Timeline & Flowchart Templates – HubSpot.pptx
SAP 2 completion done . PRESENTATION.pptx
Optimise Shopper Experiences with a Strong Data Estate.pdf
Capcut Pro Crack For PC Latest Version {Fully Unlocked 2025}
(Ali Hamza) Roll No: (F24-BSCS-1103).pptx
Factor Analysis Word Document Presentation
QUANTUM_COMPUTING_AND_ITS_POTENTIAL_APPLICATIONS[2].pptx
Database Infoormation System (DBIS).pptx
Introduction to Inferential Statistics.pptx
Leprosy and NLEP programme community medicine

5 useful things to do with click data

  • 1. 5 useful things to do with click data Dr. Gavin Ruddy gav@pontneo.com 12 Oct 2015
  • 2. 1: improve relevance match with high click rate match with low click rate boost useful itemsquery lots of items clicking provides direct user feedback
  • 3. 2: broaden responses query intersecting clickstream data extended return click traffic identifies important connections
  • 4. 3: refine point of view query focussed return e.g. filter by user type different groups can see things differently
  • 5. 4: knowledge sharing e.g. things I visit things I might like click data reflects collective understanding
  • 6. 5: implicit adaptation query dynamic return new clicking change natural change via an implicit feedback loop
  • 7. … overall … lots of items click data enables a self-organising “Knowledge Map”, built and applied from the bottom up
  • 8. … so what … • continual intelligent filtering • flexible, scalable methods that run at query time • does not require complex models, data or techniques • significant, low risk improvements • useful in many situations • effectively controlled by how you choose to integrate click data over time • helps deal with many of the usual problems associated with collaborative methods
  • 9. … but it’s complicated … • no it isn’t • click signal can be noisy, but there’s lots of it, it’s all real and it’s highly dynamic • just record appropriate clicks in a useful form for integration into your queries (e.g. fromID, toID pairs with a timestamp) • for more detail on how to go about applying these methods, see “Click-Through Filter” e.g. http://www.slideshare.net/pontneo/better-search-implementation- of-click-through-filter-as-a-query-parser-plugin-for-apache-solr- lucene
  • 10. … but it’s complicated … • no it isn’t • click signal can be noisy, but there’s lots of it, it’s all real and it’s highly dynamic • just record appropriate clicks in a useful form for integration into your queries (e.g. fromID, toID pairs with a timestamp) • for more detail on how to go about applying these methods, see “Click-Through Filter” e.g. http://www.slideshare.net/pontneo/better-search-implementation- of-click-through-filter-as-a-query-parser-plugin-for-apache-solr- lucene