Using Natural Language for Retrieval of Self-Service Business Intelligence Assets

Abstract

Ecommerce Knowledge Workers (EKW) struggle to find appropriate Self-Service Business Intelligence (SSBI) assets; Data Catalogs help here, offering the possibility to index all relevant assets from different systems and providing a search experience in a single interface. Current Data Catalogs use common metadata, usage data, social elements, curation elements and context data.SSBI assets like complex reports and interactive dashboards are often co-created by Business Intelligence- (BI-) and domain experts. The co-creation process leads to assets responding to specific stakeholder requirements and questions. During this process, many questions and sample answers are uttered. This thesis suggests to capture such utterances and storing them in metadata in order to improve Data Catalog search experience....