Paths to more personal and collaborative knowledge graphs

Paths to more personal and collaborative knowledge graphs

• 1.
  For All KGC2021Attendees Prepared by Alan Morrison Version 1.0 Paths to More Personal and Collaborative Knowledge Graphs A KGC2021 Workshop
• 2.
  Version 1.0 For AllKGC2021 Attendees Prepared by Alan Morrison Agenda Overview (Alan) Notetaking and PKGs (Ivo Velitchkov) SOLID and PKGs (Chu Nnodu) Discussion/Q&A Roam and Notetaking (Conor White-Sullivan) RDF, Notetaking and PKGs (Brian J. Rubinton) 2
• 3.
  Goal of today’sworkshop Envision a future for personal knowledge graphs (PKGs). ● PKGs are not mature--so this “workshop” won’t be a tutorial, but rather a brainstorming session. ● First question: How do we make PKGs easier for users? ● Second question: How do we get more math in the graph and not overwhelm users at the same time? ● Third question: How do we harness the momentum of the Networked Notetaking and Personal Data Protection movements? ● Hypothesis: PKGs could provide an on-ramp for users who want to help build business knowledge graphs too. Attendees are encouraged to join today’s discussion. We’ll keep things informal, so feel free to ask questions throughout. 3
• 4.
  What’s a knowledgegraph? (1 of 2) 1 2 Data (including content) that’s contextualized and disambiguated with the help of definition language and description logic 3 Rich, diverse data (including content) that’s both machine- and human-readable 4 A means of creating digital twins: Representations of people, places, things and other concepts and how they’re logically interrelated in graph form Natural language, numerical data, multimedia data and logic blended together 4 The bigger knowledge graphs (DBpedia, Wikidata, KBpedia, Blue Brain, Diffbot,Tech Sector KGs) do more of these things
• 5.
  What’s a knowledgegraph? (2 of 2) 5 A large-scale, rapid means of integration and even interoperation 7 A platform in which symbolic logic (language, reasoning and abstraction) can meld with statistical machine learning (perception and learning) 6 A means of model-driven software development 5
• 6.
  What’s a personalknowledge graph? 1 A knowledge graph for personal data (including content) 2 A means of personal data (and knowledge) management 3 A (potentially) radical journaling, creativity and productivity tool 4 A starting point for decentralizing personal data protection 6 Personal knowledge graphs are in their infancy, and a primary focus has to be on ease of use
• 7.
  Logic: A keydifferentiator in future knowledge graphs Source: Dr. Leo Obrst, Mitre; Mills Davis, Project 10X; RAI, 2010-2021 Strong Semantics Weak Semantics Quantum Logic Mathematical logic Ontology Taxonomy List Folksonomy Controlled Vocabulary Glossary Relational Model, XML DB Schema, XML Schema ER Model Topic Map RDF/S UML OWL Description Logic First Order Logic Modal Logic 2nd Order Logic Higher Order Logic Recovery Discovery Intelligence Question answering Smart Behaviors Increasing Reasoning Capability Semantic Interoperability Structural Interoperability Syntactic Interoperability Increasing Metadata, Context, & Knowledge Representation Datalog 7
• 8.
  What’s networked notetaking? [[Roam]]--> is anexample https://roamresearch.com/ 8
• 9.
  What’s SOLID (SocialLinked Data)? 1 An open means of web decentralization 2 Specs for a decentralized web API 3 Decentralized server specs and a related service (SOLID pods) that allow users to store and share their data securely 4 A starting point for decentralizing personal data protection Chu Nnodu will provide a demo and more details 9
• 10.
  Why do wecare about decentralization? ● Personal knowledge graphs need to put personal data ownership and protection first ● Users want services that work across application silos or social media walled gardens (e.g., the BBC’s music recommendation engine that works across Spotify, BBC Music, etc.) ● Users that maintain personally-identifiable information (PII) in pods or hubs can prevent duplication of their PII and verify credentials without having to share correlatable identifiers ● PII controlled and stored by users in their own pods or hubs ● To verify credentials, businesses can avoid storing PII 10
• 11.
  Why do wecare about semantics, relationships and dynamic graph data models? ● How and why things are connected is extremely important to describe context ● Ontologies, inferencing and first-order logic take us further in bridging the gap between human and machine intelligence ● Personal knowledge graphs need just enough semantics 11
• 12.
  Networked notetaking withLinked Open Data/RDF-style capabilities Kanopi.io is Brian Rubinton’s example→ And Ivo Velitchkov’s Roam augmentation is another https://kanopi.io/ 12
• 13.
  Work in Progress:A clear path to a personal knowledge graph ● A simple version of a semantic graph ● A rich means of easy integration ● A way to manage and control all sorts of personal data (including content) ● Bridges to networked notetaking and the open web ● Decentralized storage and data security that favors individual users 13 Objectives
• 14.
  For All KGC2021Attendees Prepared by Alan Morrison Version 1.0 Questions? Alan Morrison LinkedIn | Twitter | Quora | Slideshare +1 408 205 5109 [email protected] 14