Principal Data Scientist
Principal Data Scientist at LexisNexis.
Talks and Events
2022 Talk: Yes, You Can Use Knowledge Graphs In Real Life!
Developing applications that tap into knowledge graphs requires new skills and architectural patterns. In this talk, we present a typical knowledge graph system architecture, and discuss its constituent parts and functionality. We then illustrate this with a concrete example of a production system that is built to manage legal information, using Amazon Neptune, from law firms, courts, and other third-party sources. This system also offers sophisticated search functions that help lawyers find relevant information, such as similar legal briefs, to work more efficiently on their cases. The general challenge of “how to program with graphs” breaks down to questions of how to ingest, process, and consume graph data (or more generally, data that can be seen as a graph), and to do this at scale. A major goal for the overall system design is to integrate the disparate sources of data into an interconnected web of data. “New” technologies and techniques are in play, including logical reasoning and machine learning. There are also questions of modeling, such as how to enable data integration from multiple sources (e.g., “will I need an ontology for that?”), and how to support consequent question-answering based on the graph. We will also discuss how graph modeling can help in an “evolutionary” development of a system like this, through incremental additions and changes, and how thinking of the data as a product (e.g., by making it available through a reusable API) can enable new use cases and support new users. The ultimate outcome includes reusable data and a system that is “future-proof” in the face of new use cases and applications.