Aquacold - a crowdsourced query understanding and query construction tool for the linked data web

The Linked Data Web promises a disseminated, dynamic, ever expanding knowledge base where relationships between content and entities can be expressed and queried using formal logic rules. The data formats and exchange protocols that comprise the Linked Data Web have existed for over 15 years and grown to incorporate over 149 billion triples but there is no established 'mainstream' system that allows regular non-technical users to query the linked data web using natural language. In this paper we introduce Aquacold (Aggregated Query Understanding And Construction Over Linked Data), a novel Linked Data query tool designed to fill this gap. Aquacold combines a simple browsing interface for exploring, filtering and sorting linked data with query labelling which allows the user to store a natural language representation of the underlying SPARQL query. A search interface allows users to write guided natural language queries, which are converted into SPARQL queries to retrieve results from the linked data source. A set of voting tools are presented for each query and associated result set to surface the most accurate templates and relegate those that are less accurate. This crowdsourced approach for labelling, saving and voting on linked data query templates enables regular, non-technical users to make complex natural language queries over the linked data web without having to use a structured language such as SPARQL. In this paper we provide an overview of the system and discuss an early prototype.