🚪Navigating the Nodes: Tips on Building Your Knowledge Graph
Frontdoor Finds - Issue 11
William Mathews & Frontdoor
July 31, 2023
As builders, we spend a lot of time consuming knowledge on the web. At Frontdoor, we've built a personal librarian for the internet to help. Each fortnight we pick a topic, and share the most curated tweets, blogs and books on Frontdoor....
3️⃣ Tweets
1/ Raymond D Sims, a Great Thread on Knowledge Management Resources
I fell into the #KnowledgeGraph (KG) and Personal Knowledge Graph (PKG) rabbit hole. After too many hours of reading and watching, I'm now smarter than before; however, I am still learning.
Some stops along the way:
⬇️— Raymond D Sims (@rsims)
Aug 12, 2022
2/ Raymond D Sims on Useful Concepts in Knowledge Graphs
Useful concepts in @onotext
🔹KG use data management Database, Graph & Knowledgebase paradigms
🔹RDF representation provides Expressivity, Performance, Interoperability & Standardization
🔹Modeling w/ Classes, Relationship Types, Categories & Descriptions— Raymond D Sims (@rsims)
Aug 12, 2022
3/ Towards Data Science on Creating Knowledge Graphs
. @SteveHedden offers a step-by-step guide to creating a knowledge graph and exploring its potential to enhance an LLM.
— Towards Data Science (@TDataScience)
Jul 25, 2023
2️⃣ Articles
1/ Lisa Ehrlinger and Wolfram Wöß on the Definition of Knowledge Graphs
🤖 Frontdoor Summary:
Despite being frequently used in research and business, there's no universally accepted definition for "knowledge graph". Interpretations vary widely, often leading to contradictions.
The understanding and perception of knowledge graphs have been significantly influenced by Google's Knowledge Graph and Wikipedia's descriptions.
Knowledge graphs differ from related concepts like knowledge bases and ontologies primarily due to their ability to integrate information and employ a reasoning engine.
It is proposed that a knowledge graph should be defined as a structure that acquires and integrates information into an ontology and utilises a reasoner to derive new knowledge. It's also clarified that the size of a knowledge graph is not a defining characteristic, but rather the focus should be on its reasoning capabilities and the integration of information sources.
2/ Building Knowledge Graphs
🤖Frontdoor Summary:
This is a great practical guide to building an academic knowledge graph with OpenAI and a graph databases. It give detailed instructions on how to:
Use the Arxiv API to fetch metadata of papers and save it into the knowledge graph: This point is essential because it describes the data source and how it's gathered.
Use GPT-3 to extract entities and relationships from the title and summary of papers: This point explains how AI is used to automatically extract relevant information from the papers, which is key to building the knowledge graph.
Create constraints for uniqueness in the graph database: This point is important for maintaining the integrity of the data in the graph database.
Use GPT-3 to generate Cypher statements for querying the knowledge graph: This point illustrates the application of AI to not just building, but also querying the knowledge graph. This could be a significant advantage in terms of ease and flexibility of use.
1️⃣ Book
Ivo Veltchikov and George Anadiotis on personal knowledge graphs
🤖Frontdoor Summary:
Knowledge Graphs (KGs) depict relationships between real-world entities, aiding search engines, social networks, and more.
Personal Knowledge Graphs (PKGs) are individualized KGs, encoding a user's personal data and common-sense knowledge.
PKGs fuel personalized digital applications like coaches and health trackers, enabling more effective decision-making.
The book explores advanced PKG research, including construction approaches, personalization modeling, evaluation methods, and knowledge representation.
The content is aimed at professionals and students in data science, ICTs, knowledge engineering, semantic web, and machine learning.
Was this newsletter forwarded to you? Sign up Here