Customer Intelligence on Social
Media
Rosaria Silipo
KNIME.com AG
Copyright © 2015 KNIME.com AG
Copyright © 2014 KNIME.com AG
2
The KNIME Platform: Open for Innovation
Powerful:
Legacy  Future Tools
Collaborative:
Scientists  Analysts
Integrative:
Legacy  Future Data
Transparent:
Existing  Future Expertise
Agile:
Internal  External Wisdom
Copyright © 2014 KNIME.com AG
3
3
The KNIME Analytics Platform
Copyright © 2014 KNIME.com AG
4
4
Over 1000 native and embedded nodes included:
MySQL, Oracle, etc.
SAS, SPSS, etc.
Excel, Flat, etc.
Hive etc.
XML, PMML
Text, Doc, Image
Web Crawlers
Industry Specific
Community / 3rd
Copyright © 2014 KNIME.com AG
ETL
Row,
Column
Matrix
Text, Image
Time Series
Java
Python
Community / 3rd
Statistics
Data Mining
Machine Learning
Web Analytics
Text Mining
Network Analysis
Social Media Analysis
WEKA
R
JFreeChart
Community / 3rd
5
5
via BIRT
PMML
XML
Databases
Excel, Flat, etc.
Hive etc.
Spark
Text, Doc, Image
Industry Specific
Community / 3rd
Copyright © 2014 KNIME.com AG
11
The Problem
A major European Telco
Its Forum Site
• Can you tell us what people say about our new
product?
• Can you tell us who is supporting the product and
who trashing it?
• Of those, can you tell us who is an influencer?
Copyright © 2014 KNIME.com AG
12
12
The Data
• The Data Set unfortunately cannot be shared
• Slashdot Forum Data are!
• Slashdot was a public forum built in 1997 and
hosting a number of discussions: from software to
philosophy, from science fiction to politics.
• Politics was the biggest discussion group
• So, politics is what we analyzed to find out:
– What users were thinking about a political issue
– Who was pro and who was con
– Who was an influencer
Copyright © 2014 KNIME.com AG
13
13
Copyright © 2014 KNIME.com AG
14
The Politics Group in the Slashdot DataSet
•
•
•
•
24 000 non anonymous users
496 posts
140 000 comments
Most posts have around 200 comments
Copyright © 2014 KNIME.com AG
15
15
Copyright © 2014 KNIME.com AG
16
Text Analytics: Options
•
•
•
•
Tag (Word) Clouds
Topic Detection
Topic Shift
Sentiment Analysis
I find PRODUCT X to be very good and useful,
but it is a bit too expensive.
Copyright © 2014 KNIME.com AG
17
Copyright © 2014 KNIME.com AG
Scatter Plots
and Tag Clouds
Sum of Sum of
frequencies of positive
and negative words per
user
Sum of frequencies
of positive and
negative words per
post/comment
Read Data
Text Analytics: Workflow
Document type
is required
18
Loading MPQA
Stanford dictionary for
sentiment attribute
Text Analytics: Results
Most positive
and most
talkative user
dada21
Most negative
user pNutz
Copyright © 2014 KNIME.com AG
19
Text Analytics: Open Questions
• Is dada21 an influencer?
• Is pNutz an influencer?
• Shall we take marketing actions about any of
them?
Copyright © 2014 KNIME.com AG
20
20
Copyright © 2014 KNIME.com AG
21
Network Mining: Options
• User Interaction Graph
• Influencers vs. Followers
• User Network Investigation
Copyright © 2014 KNIME.com AG
22
Network Mining: Workflow
Create Empty
Network
Scatter Plots
Centrality Index for
authority score
Read Data
Create
Network
Content
Copyright © 2014 KNIME.com AG
Extract
largest
sub-graph
23
Network Mining: Results
Dada21
Carl Bialik from the WSJ
Doc Ruby
Copyright © 2014 KNIME.com AG
24
Network Mining: Open Questions
• Is Carl Bialik from WSJ positive or negative about the
topic?
• Is dada21 positive or negative about the topic?
• What shall we do marketing-wise with noninfluencers such as doc Ruby?
Copyright © 2014 KNIME.com AG
25
25
Copyright © 2014 KNIME.com AG
26
Text Analytics and Network Mining: Workflow
Network Mining
Scatter Plots
Read Data
Joiner
Sentiment Analysis
Copyright © 2014 KNIME.com AG
27
Text Analytics and Network Mining: Results
dada21
Carl Bialik from the WSJ
Tube Steak
WebHosting Guy
Catbeller
Doc Ruby
99BottlesOfBeerInMyF
pNutz
Copyright © 2014 KNIME.com AG
28
Text Analytics and Network Mining: Results
Copyright © 2014 KNIME.com AG
29
Copyright © 2014 KNIME.com AG
30
Conclusions
• Is Carl Bialik from WSJ is an influencer and …
neutral.
• Most influencers are actually neutral.
• Worth it keep informed
• dada21 is talking positively about each topic. Worth
it to pamper him/her.
• Of the negative talking users, pNutz though
obnoxious, is not the main worry. Catbeller is.
Copyright © 2014 KNIME.com AG
31
31
Where can I find all this?
White paper, Workflows, and Data is available on the
KNIME web site:
http://www.knime.com/white-papers (section Social
Media)
https://www.knime.org/files/knime_social_media_whit
e_paper.pdf
Copyright © 2014 KNIME.com AG
32
Resources
•
KNIME (www.knime.org)
•
•
•
•
•
•
•
BLOG for news, tips and tricks(www.knime.org/blog)
FORUM for questions and answers (tech.knime.org/forum)
EXAMPLE SERVER for example workflows
LEARNING HUB (www.knime.org/learning-hub)
KNIME TV channel on
KNIME on
@KNIME
KNIME on
https://www.facebook.com/KNIMEanalytics
Copyright © 2014 KNIME.com AG
33
33
Thank You
Free Copy of KNIME Beginner’s Luck Book
at KNIME Press
https://www.knime.org/knimepress
Promotion Code:
Barcelona2016
Copyright © 2014 KNIME.com AG
34