Vector Databases 101 - An introduction to the world of Vector Databases
Generating actionable consumer insights from analytics
1. Generating actionable consumer
insights from analytics
Shazri,
Researcher , Advanced Informatics,TM R&D
Researcher ,biophotonics group, photonics research centre, univ. malaya
mshazri@gmail.com
2. talk roadmap
information & data sources trends
challenges and roadblocks in big data and
analytics implementation
opportunities too
use cases in health and telecommunication
3. *Digression: market <-> user
spectrum.
Traditional Survey
Neo-Historical
Based Analysis
Say emotion/
satisfaction + neo-
historical analysis
context Action unto
Market wide
Range of
time
Customer
Specific
Right now
Customer
Specific +
Using @ wire
time info
Right now
Note: more variables , unto shorter time to action.
4. **information and data sources
trend
Personalized marketing; more powerful tabs -
increased data vol. per person.
Serv. tech unto satisfaction; more real time
network elements reporting
Emotion detection unto reaction/action;
unstructured and structured relational
sources
6. challenges and roadblocks in big data and
analytics implementation
different db sources have different
owners, - to target higher level of
analytics abstraction.
interfacing with unstructured
sources speed
demand of output speed
security ?
7. Zoom in - Security
The following lists the security challenges in Big Data. The list was taken from Top Ten Big Data Security and
Privacy challenges, by Cloud Security Alliance.
• Secure computations in distributed programming frameworks
-Untrustworthy data mapper.
• Security best practices for non-relational data stores
-No in database security yet, now depends on middleware.
• Secure data storage and transactions logs
-No tier-ing strategies to differentiate type of data.
• End-point input validation/filtering
-Veracity, how do you make sure data is trustworthy.
-many factor validation/verification.
• Real-time security/compliance monitoring
-what is anomalous in Big Data framework.
-baseline signature?
• Scalable and composable privacy-preserving data mining and analytics
-some analytics results can be correlated with other external results that can infer identity.
• Cryptographically enforced access control and secure communication
-access must be encrypted.
• Granular access control
-Giving access to the right people, without being too troublesome.
• Granular audits
-What happened? When?
• Data provenance
-Meta data security and speed of processing.
8. opportunities too
md:more organic and whole measures map to
objective function
nw:richer outputs
nw:high velocity outputs
lg:cross web-app , mobile-app , local & secured
sources search.
9. use cases in health and
telecommunication
t:call centre - customer care and management
h:genomic analysis
t:high performance analytics and security