Semantic Table Retrieval Using Keyword and Table Queries

Tables on the Web contain a vast amount of knowledge in structured form. To tap into this valuable resource, we address problem table retrieval: answering an information need with ranked list tables. We investigate two different variants, based how is expressed: as keyword query or existing (“query-by-table”). The main novel contribution work semantic retrieval framework for matching needs (keyword queries) against Specifically, (i) represent queries and tables multiple spaces (both discrete sparse continuous dense vector representations) (ii) introduce various similarity measures those representations. consider all possible combinations representations use these features supervised learning model. Using purpose-built test collections Wikipedia tables, demonstrate significant substantial improvements over state-of-the-art baselines.