sbsgrid in_memory-reasoner_and_search_query_engine
DESCRIPTION
TRANSCRIPT
© SBSGRID.NET
SBSGRID
IN-MEMORY REASONING & SEARCHQUERY ENGINE
SBS Semantic Business Solutions, Inc
Overview
© SBSGRID.NET
INTRODUCTION LINKED-DATA
Linked-Data is a substantial movement & technology for:
Meaning Based Computing
Next Generation Search & Query
Next Generation Information Discovery
Web based Machine Learning & AI
1 see http://www.linkeddata.org
1
© SBSGRID.NET
INTRODUCTION LINKED-DATA
Linked-Data is also the enabler for the so-called Web 3.0
1 see http://www.linkeddata.org
1
Linked-Data approach works (vs. SemWeb)
© SBSGRID.NET
INTRODUCTION LINKED-DATA
Linked-Data are virtualizing the data-source:
STRUCTURED
LINKED-DATA
RelationalDBs Wikis Documents Websites
SEMISTRUCTURED
UN-STRUCTURED
Wrapper Wrapper Crawler Crawler
Mediated Schema
© SBSGRID.NET
INTRODUCTION SEARCHQUERY
SearchQueries leverage the power of Linked-Data:
Contextual & Associative Queries
On Data Level (No Semantic Search)
Information Travel from any Angle
Question in Free Form
© SBSGRID.NET
INTRODUCTION SEARCHQUERY
SearchQueries operate on interlinked data (not only linked concepts):
Complex Queries (any # of keywords !!)
Conditions (>, <, =, etc)
Semantic Operators (“today”<->date info)
Analytics (sum, sqrt, div, etc)
© SBSGRID.NET
SearchQuery: Quick Tour
Unlinked Data (Relational Database)
Dataset example from the art world
INTRODUCTION SEARCHQUERY
© SBSGRID.NET
Linked-Data (Semantic FactBase)
INTRODUCTION SEARCHQUERY
SearchQuery: Quick Tour
Linked Concepts + Linked Data
© SBSGRID.NET
"Galleries in NYC selling Modern Painting below 8k"
Any # of Keywords are Contextually Matched
INTRODUCTION SEARCHQUERY
SearchQuery: Quick Tour
© SBSGRID.NET
“Current Exhibitions on Gerhard Richter in Germany"
The Semantic Operator “Current Exhibitions” is matched against Date Information
Note the potential Polymorphic Role on “Gerhard”
INTRODUCTION SEARCHQUERY
SearchQuery: Quick Tour
© SBSGRID.NET
“Japanese and Korean Art between 10k & 20k"
The Price Range executes > and < Operations
INTRODUCTION SEARCHQUERY
SearchQuery: Quick Tour
© SBSGRID.NET
Applicability
SBSGRID Application Scenarios…
© SBSGRID.NET
APPLICABILITY
1 see Slides: SBSGRID SearchQuery Explained
2 see White Pager: Visual Linked-Data Publishing
Data Source A
Wrapper
Scenario: One Database
In-MemorySBSGRID Engine
Browser SearchQuery
High Performance
Automatic Semantics2
1
Self-Governed
© SBSGRID.NET
Scenario: Many Databases
Browser SearchQuery• Works Instant & Incremental
• Contains Ontology Alignment
• Meaning based Information Linking
1 see White Paper: Meaning based Information Linking
1
Data Source A
Wrapper
Data Source A
Wrapper
Data Source A
Wrapper
In-MemorySBSGRID Engine
Self-Governed
APPLICABILITY
© SBSGRID.NET
Data Source A
Data Source A
Data Source A
ETLData
Warehouse
Classic Relational Approach
Review: ETL Process
APPLICABILITY
© SBSGRID.NET
Scenario: Contextual Analytics
Contextual Cubes on the Fly
Browser
In-MemorySBSGRID Engine
1 see White Paper: SBSGRID Contextual Analytics
1
APPLICABILITY
© SBSGRID.NET
Scenario: Linked-Data Publishing
1 see White Pager: Visual Linked-Data Publishing
Data Source A
Wrapper
In-MemorySBSGRID Engine
Browser Visual Process1
Self-Governed
• Drag & Drop Interface
• Instant SearchQuery
• JDBC/ODBC Data-sourcesUI Tool
APPLICABILITY
© SBSGRID.NET
Scenario: Triple-Store Caching
TripleStore
In-MemorySBSGRID Engine
Browser SearchQuery
High Performance
1
• Transactional Store
• RDF/RDFS/OWL Support
• SPARQL Support
APPLICABILITY
© SBSGRID.NET
Scenario: NLP Leverage
APPLICABILITY
Browser SearchQuery• Works Instant & Incremental
• Contains Ontology Alignment
• Meaning based Information Linking
1 see White Paper: Meaning based Information Linking
1
WIKI
NLP
Documents
NLP
Websites
NLP
In-MemorySBSGRID Engine
© SBSGRID.NET
SUMMARY
SBSGRID is the first tool for Dynamic Linked-Data:
No Ontology Development
No Middleware Development
Visual Linked-Data Publishing
Instant & Incremental Use
© SBSGRID.NET
Appendix
Some more info’s…
© SBSGRID.NET
APPENDIX
• HPC Engine
• Based on Java, TCP/IP, XML & HTML5
• Designed for Pure In-Memory (Compressed Index / Data)
• Hybrid of GraphQuery/FullTextSearch/Cache Engine
• Reasoner is aware about structural graph characteristics
• Optimized structures for syntactic & semantic decomposition
• WEB API
• XML-REST API
• JavaScript Classes for Continuous Query
• Java & C# API
• SPARQL++ Functionality
• Non Transactional (Compensate Action)
© SBSGRID.NET
APPENDIX
• Import API
• Triple Strings TCP/IP Interface
• Java / C# Client Interface
• Hot Updates (incremental index creation)
• Bulk Load
• Small Code
• No usage of libraries (no context switch; high security)
• 100% ownership (high end optimization)
• Specific Reasoner requests can be implemented
© SBSGRID.NET
APPENDIX
• Queries
• N-dimensional Reasoner- covers all query cases wanted- covers combinations of concepts, terms, data & operators
• Any Number of Keywords (still precise results)
• Support for Ranks, Synonyms, Operators
• Context Aware Full-Text-Server
• User Escort (Semantic Suggest Lists, Context Maps)
© SBSGRID.NET
APPENDIX
• Deployment
• JAR Deployment (with our without servlet container)
• No Middleware Development- Semantic Queries are created automatically- Data Deployment on fact level (config vs. coding)
• Data publishing and import through visual tool
• Data transport via TCP/IP
• Embedded Web-Server (optional)
• Admin GUI (runtime statistics)
• Security
• Access entitlement on field, object or namespace level
• No usage of external libraries (high security, 100% ownership)