View
4.820
Download
2
Category
Preview:
DESCRIPTION
Vital.AI's Big Data App Platform allows creating data-driven apps rapidly.
Citation preview
C r e a t i n g I n t e l l i g e n t A p p sw i t h S e m a n t i c s & B i g D a t a
O c t o b e r 2 , 2 0 1 3
Marc C. Hadfield, FounderEmail: marc@vital.ai
Thursday, October 3, 13
<introduction>Marc HadfieldFounder of Vital.AI
Vital.AIBig Data Application Platform:
Semantics & Big Data in one platform,
combining multiple Data Analysis
techniques.
Thursday, October 3, 13
Intelligent Applications
Development Processes
Business Value
Today:
Thursday, October 3, 13
Intelligent Applications:
...learn from experience.
Thus, are experience-driven: data-driven.
“self-optimizing”
Thursday, October 3, 13
Data-Driven Applications
Thursday, October 3, 13
Moderni.st
Thursday, October 3, 13
Moderni.st
Thursday, October 3, 13
Personal Agent App
Thursday, October 3, 13
Why are data-driven applications different?
Thursday, October 3, 13
User Interface
Business Logic
Database
Traditional Applications
Why are data-driven applications different?
“sta%c”
Thursday, October 3, 13
Why are data-driven applications different?
“dynamic”
User Interface
Business Logic
Events
Data-Driven Applications
Data
Analytics
Derived Behavior
Logic
Thursday, October 3, 13
Data-Driven Applications use many different technologies, with many different skill sets:greater power, higher cost.
Thursday, October 3, 13
Why do we want a data-driven application?
Thursday, October 3, 13
What are the components of a data-driven application?
Data Collection
Dynamically Adjust Behaviors
Data Analysis / Feedback
Thursday, October 3, 13
Implementing a Data-Driven Application.
Thursday, October 3, 13
Implementation Goals:
Off the shelf?
Minimize Impedance Mismatch:Mobile/Web User Interface vs.
Semantic vs.Big Data vs.
...
Efficient Development Processes
Thursday, October 3, 13
Vital.AI Platform Stack
Clean separation of responsibilities and skill sets.
Rapid Development of Data-Driven Apps.
VitalSigns: Ontology-based Data Model.
Thursday, October 3, 13
Vital Prime
REST Interface
Collects Real-Time Events
Database Interfaces:HBase, Allegrograph, ...
Workflow Interface
Script Engine
In-Memory Analytics
User State Management
Thursday, October 3, 13
Vital Flows
Implements Workflows:
Natural Language Processing
Run Predictive Analytics
Graph Analytics
Logical Inference
Many open-source components in a common workflow framework.
Thursday, October 3, 13
Hadoop
Implements Big Data Analysis:
Machine Learning
Build Predictive Models
Thursday, October 3, 13
Vital Core Ontology
Thursday, October 3, 13
Vital Core Ontology
Vital Domain Ontology
Application Domain Ontology
Extending the Ontology
Thursday, October 3, 13
Generating Data Bindings with VitalSigns:
Ontology VitalSigns
Groovy Bindings
Semantic Bindings
Hadoop Bindings
Prolog Bindings
Graph Bindings
HBase Bindings
JavaScript Bindings
Thursday, October 3, 13
person123.name = "John"person123.worksFor.company456
<person123> <hasName> "John"<worksFor123> <hasSource> <person123><worksFor123> <hasDestination> <company456><worksFor123> <hasType> <worksFor>
person123, Node:type=Person, Node:hasName="John"worksFor123, Edge:type=worksFor, Edge:hasSource=person123, Edge:hasDestination=company456
Groovy
RDF
HBase
Data Representations
Thursday, October 3, 13
Editing the Ontology
Thursday, October 3, 13
Developing with the Ontology in UI, Hadoop, NLP, Scripts, ...
Node:Person Node:PersonEdge:hasFriend
Set<Friend> person123.getFriends()
“Best Practices” in Ontology Development.
Thursday, October 3, 13
Data Analysis
Thursday, October 3, 13
Using Natural Language Processing
Topic Categorization Extract Entities
Dialogue System Entity Normalization
Thursday, October 3, 13
Using Graph Analytics
PageRank, Centrality, Interest Graph, ...
Thursday, October 3, 13
Using Machine Learning
Implemented via Hadoop
Algorithms from Mahout
Build Predictive Models
Models used in Workflows
Data defined in Ontology
Classification, Clustering...
Thursday, October 3, 13
Data Visualization - Cytoscape
Thursday, October 3, 13
Data Management - Dashboard, Wiki
Thursday, October 3, 13
Examples
Shopping Recommendation App
Personal Agent App
Content Recommendation App
Thursday, October 3, 13
Content Recommendation
REST Interface
Vital Client
Vital Prime
Vital Flow Queue
Integrator NLP HadoopML Model
S3
Elastic Map/Reduce
Mahout
300M Users
100K Publishers
NLP
Machine Learning
Content + User Signals
Recommendations
Thursday, October 3, 13
Personal Agent
Dialogue System
Thursday, October 3, 13
Personal Agent
Interest Graph Recommendations
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Moderni.st
App Server
Vital Prime
IntegratorFlow
NLP Flow Logger Flow
Inference Flow
VITAL APIVitalSigns
Web Browser /Mobile Device
HDFS
Hbase Hadoop JobsMahout
Vital Flows
Flume
Allegrograph
Thursday, October 3, 13
Creating an Intelligent App
Create User Interface.Use Vital Client, Send Signals
Create App Ontology, includingDomain Classes & Events/Signals
Define Workflows for Data Processing
Define Predictive Models, Build Models
Create scripts for App Functionality
Deploy Components
Thursday, October 3, 13
Development Processes
Use the right tool for the right job,
& the right people for the right job.
Integrated Framework.
Coordinate on Ontology.
Eliminate Data Mis-Matches.
Clean code separation of “paradigms.”
Solve “Variety” Big Data challenge.
Thursday, October 3, 13
Business Value
Intelligent App.
Efficient, Rapid Development.
Flexibility to learn, adapt, improve.
Greatly reduce risk.
Thursday, October 3, 13
For more information, please contact:Marc C. Hadfield, FounderE: marc@vital.aiP: 917.463.4776
51
THANK YOU!
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Thursday, October 3, 13
Recommended