SLA 2024 Annual Conference & Expo

Taxonomies are as relevant as ever. Search technologies (matching words in texts) still have accuracy limitations, users have come to expect an like faceted search, wider adoption of linked (open) data relies on shared metadata is based on controlled vocabularies, the connected data benefits within enterprises of ontologies and knowledge graphs depend on underlying taxonomies, personalization and recommendation also rely on taxonomies, and even the newest large language models yield improved results when provided with context coming from a taxonomy. To be effective, taxonomies need to be custom-designed to fit the context of the content and users involved in a particular implementation. Many organizations have taxonomies already, but they likely were created for limited, legacy purposes. New initiatives and new use cases require new or renewed taxonomies. Legacy taxonomies are just one of several sources to consider when creating a new taxonomy. This presentation will take a fresh look at designing and creating taxonomies, considering the latest methods recommended by taxonomy consultants. Taxonomies are built with a combination of bottom-up and top-down methods. Bottom-up includes manual analysis and using various technology solutions (including text analysis, search log analysis, and generative AI). Top-down methods focus on the collaborative input of stakeholders (including workshops, focus groups, and interviews). Collaboration is also needed to maintain a taxonomy. Whether you are new to creating taxonomies or would like to know the latest recommendations, this session will provide valuable recommendations and tips to get you started.