Member-only story

From Raw Text to Wikidata Taxonomy and Knowledge Graph

Extracting knowledge semi-automatically with GCP, GPT, and Wikidata

Sixing Huang
CodeX
10 min readAug 22, 2023

--

Photo by Sincerely Media on Unsplash

A knowledge graph is a graph that captures relationships between entities, concepts, and facts in a specific domain. It consists of nodes (entities or concepts) and edges (relationships between nodes). We can use them to visualize complex relationships in the real world, such as the buyers and sellers in the business world. We can also superimpose taxonomies into the graph and make it even more powerful. For example, if the first supplier is no longer available in a supply chain, we can quickly find the next best alternative and redirect the orders.

At first glance, navigating a knowledge graph might seem complex, but it’s deceptively easy thanks to no-code tools like Gemini Data and chatbots, whereas building one seems uncomplicated but presents the true hurdle. It is particularly so when you start with raw texts. At a high level, the creation of a knowledge graph requires a combination of programming skills, linguistics, and domain knowledge. At the programming level, natural language processing (NLP) involves named entity recognition (NER), named-entity linking (NEL), relationship extraction, and ontology modeling. Different domains or applications have their specific vocabularies and ontologies and require the…

--

--

CodeX
CodeX

Published in CodeX

Everything connected with Tech & Code. Follow to join our 1M+ monthly readers

Sixing Huang
Sixing Huang

Written by Sixing Huang

A Neo4j Ninja, German bioinformatician in Gemini Data. I like to try things: Cloud, ML, satellite imagery, Japanese, plants, and travel the world.

Responses (1)