Associative Semiotic
Hypergraph Database
We build powerful meaningful relationships easily
In Intersystems Cache with Python Pandas
By Athanassios I. Hatzis, PhD
http://healis.eu/about
Overview
•Our focus on the “Big” Data problem
•Our approach: A Multi-Perspective
Database Framework
•Demo of key differentiating factors
– The mapping problem
– Filtering mechanism
– Traversal
•Our objectives
•Our future plan
16/11/2017 Athanassios I. Hatzis, PhD 2
Integration of Data Resources
•4V’s of Integration
– Veracity (Certainty of data)
– Validity (Cleansing of data)
– Variety (Diversity of data, Container)
– Variability (Semantics of data, Content)
•Extract, Transform, Load
•Enrichment
•Correlation and Aggregation
•Retrieval mechanism
16/11/2017 Athanassios I. Hatzis, PhD 3
Time for Smart Data
•A Multi-Perspective Framework for
Multi-Model Databases
– Conceptual (Entities/Attributes)
– Logical (Hypergraph)
– Data Model (ER, Topic Maps, RDF, XML/JSON)
– Physical (Key-Value Store)
– Semantic (Triangle of Reference)
– Object-Oriented (TBox, ABox)
– Programming (Functional)
– Mathematical (Sets and Morphisms)
16/11/2017 Athanassios I. Hatzis, PhD 4
Associative Semiotic
Hypergraph• Associative In the core
– Associative Multidimensional Arrays
– Associative Entity
• Semiotic By Design (R3/S3)
– Resources, Representation, Realization
– Signified, Sign, Signifier
– Model, Identifier, Data Values
• Hypergraph In Logic
– Hyperedge (HE – connects many hypernodes)
– Hypernode (HN – is always connected to HE)
– Hyperlink (HL – connects a pair of HE, HN)
16/11/2017 Athanassios I. Hatzis, PhD 5
Hypergraph
• Structure
– Hyperedge (HE – connects many hypernodes)
– Hypernode (HN – is always connected to HE)
– Hyperlink (HL – connects a pair of HE, HN)
• Hyperedge Represents
– Entities, Records (Hyperbond), Tables, etc…
– One side in a Many to Many relationship
• Hypernode Represents
– Attributes, Values (Hyperatoms), Columns, etc…
– One side in a Many to Many relationship
• Hypelink Represents
– A Many to Many Binary Relationship (e.g. Supplier – Part)
• Directed, Inverse, Undirected
– Bidirectional Edge that connects a tail node to a head node
16/11/2017 Athanassios I. Hatzis, PhD 6
Hypergraph vs Graph
•Why Hypergraph
– Natively closer to n-ary Relation
– Entity (Relation)  Hyperedge
– Attribute Instance  Hypernode
– Attribute Domain (Set)  Hypercollection
– Association  (Tuple or Entity Relationship)
– Relational DBMS Reference DBMS
•Advantages
– Easier navigation
– Better fine-grain control
16/11/2017 Athanassios I. Hatzis, PhD 7
Demo Mapping
16/11/2017 Athanassios I. Hatzis, PhD 8
Demo Filtering
16/11/2017 Athanassios I. Hatzis, PhD 9
Demo Traversal
16/11/2017 Athanassios I. Hatzis, PhD 10
Our Objectives
•Facilitate collaboration across the
Python and Intersystems Cache
developers communities
•Promote our technology to developer
communities
•Act as a bridge between vendors, their
partners and industry clients
•Distribute and apply our technology to
vendors and their customers
16/11/2017 Athanassios I. Hatzis, PhD 11
Our Future Plan
•Development and testing of a
Minimum Viable Product
•Collaborate with our partners to build
a self-service associative, semiotic BI
system
•Make the technology open and
accessible to the general public
•Promote and apply both proprietary
and Open Source implementations
16/11/2017 Athanassios I. Hatzis, PhD 12
Our Mantra
•We build powerful meaningful
relationships easily
•Acknowledgements
•Thank you
16/11/2017 Athanassios I. Hatzis, PhD 13

Athanassios Hatzis

  • 1.
    Associative Semiotic Hypergraph Database Webuild powerful meaningful relationships easily In Intersystems Cache with Python Pandas By Athanassios I. Hatzis, PhD http://healis.eu/about
  • 2.
    Overview •Our focus onthe “Big” Data problem •Our approach: A Multi-Perspective Database Framework •Demo of key differentiating factors – The mapping problem – Filtering mechanism – Traversal •Our objectives •Our future plan 16/11/2017 Athanassios I. Hatzis, PhD 2
  • 3.
    Integration of DataResources •4V’s of Integration – Veracity (Certainty of data) – Validity (Cleansing of data) – Variety (Diversity of data, Container) – Variability (Semantics of data, Content) •Extract, Transform, Load •Enrichment •Correlation and Aggregation •Retrieval mechanism 16/11/2017 Athanassios I. Hatzis, PhD 3
  • 4.
    Time for SmartData •A Multi-Perspective Framework for Multi-Model Databases – Conceptual (Entities/Attributes) – Logical (Hypergraph) – Data Model (ER, Topic Maps, RDF, XML/JSON) – Physical (Key-Value Store) – Semantic (Triangle of Reference) – Object-Oriented (TBox, ABox) – Programming (Functional) – Mathematical (Sets and Morphisms) 16/11/2017 Athanassios I. Hatzis, PhD 4
  • 5.
    Associative Semiotic Hypergraph• AssociativeIn the core – Associative Multidimensional Arrays – Associative Entity • Semiotic By Design (R3/S3) – Resources, Representation, Realization – Signified, Sign, Signifier – Model, Identifier, Data Values • Hypergraph In Logic – Hyperedge (HE – connects many hypernodes) – Hypernode (HN – is always connected to HE) – Hyperlink (HL – connects a pair of HE, HN) 16/11/2017 Athanassios I. Hatzis, PhD 5
  • 6.
    Hypergraph • Structure – Hyperedge(HE – connects many hypernodes) – Hypernode (HN – is always connected to HE) – Hyperlink (HL – connects a pair of HE, HN) • Hyperedge Represents – Entities, Records (Hyperbond), Tables, etc… – One side in a Many to Many relationship • Hypernode Represents – Attributes, Values (Hyperatoms), Columns, etc… – One side in a Many to Many relationship • Hypelink Represents – A Many to Many Binary Relationship (e.g. Supplier – Part) • Directed, Inverse, Undirected – Bidirectional Edge that connects a tail node to a head node 16/11/2017 Athanassios I. Hatzis, PhD 6
  • 7.
    Hypergraph vs Graph •WhyHypergraph – Natively closer to n-ary Relation – Entity (Relation)  Hyperedge – Attribute Instance  Hypernode – Attribute Domain (Set)  Hypercollection – Association  (Tuple or Entity Relationship) – Relational DBMS Reference DBMS •Advantages – Easier navigation – Better fine-grain control 16/11/2017 Athanassios I. Hatzis, PhD 7
  • 8.
  • 9.
  • 10.
  • 11.
    Our Objectives •Facilitate collaborationacross the Python and Intersystems Cache developers communities •Promote our technology to developer communities •Act as a bridge between vendors, their partners and industry clients •Distribute and apply our technology to vendors and their customers 16/11/2017 Athanassios I. Hatzis, PhD 11
  • 12.
    Our Future Plan •Developmentand testing of a Minimum Viable Product •Collaborate with our partners to build a self-service associative, semiotic BI system •Make the technology open and accessible to the general public •Promote and apply both proprietary and Open Source implementations 16/11/2017 Athanassios I. Hatzis, PhD 12
  • 13.
    Our Mantra •We buildpowerful meaningful relationships easily •Acknowledgements •Thank you 16/11/2017 Athanassios I. Hatzis, PhD 13

Editor's Notes

  • #10 Filtering Data Sets (Collections) Filtering Tuples We are NOT traversing the graph based on the labels of the edge, We are FILTERING and ADDRESSING its constructs (Hypernodes, Hyperedges, Hyperlinks)