Prashanth Rao, Kùzu, Inc, AI Engineer
Paco Nathan, Senzing, Principal DevRel Engineer

An important obstacle for adopting knowledge graph technology in enterprises is that virtually all of the enterprise-level data is originally stored in unstructured formats or in some structured but non-graph format, such as tables. Therefore, an inevitable first step in adopting graph technologies is to convert these data sources into a high-quality KG.

This tutorial walks through the different steps of this process and covers a suite of tools and technologies one can use. Broadly, KG construction can be divided into three high-level steps: (1) Unstructured text to basic knowledge graph construction: incl. text parsing using NLP libraries, creating lexical and text graphs, named entity/relationship extraction, and entity ranking. (2) Linking the KG with existing structured data such as entities and relationships with those in the basic knowledge graph. (3) KG quality enhancement, which involves steps like entity resolution to increase the quality or specificity of the extracted entities and relationships.

The tutorial will demo on a live example the evaluation of KGs as it is transformed between these steps that uses several open-source and commercial technologies, such as Senzing, Kùzu (an embedded graph database), and several alternative tools that can be used.

Prerequisites:

Some experience coding in Python
Familiarity with popular packages such as Pandas, Jupyter, Docker
Intructions to participate:

Make sure to have Python, Git, and Docker installed on your laptop in advance.
Then run one command:
docker pull senzing/demo-senzing