Supervisors info:
Εμμανουήλ Κουμπαράκης, Καθηγητής, Τμήμα Πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
Summary:
Question Answering over knowledge graphs has been studied a lot in recent years. The main aspect of such systems is to provide an interface, through which natural language questions can be posed and answered. Said systems generate queries and retrieve data from knowledge bases, usually in URI form. Thus, it is important to present this information appropriately, so that any user can make sense of the answers. We have developed an interface to the GeoQA system, which is a question answering engine over linked geospatial data. The interface of GeoQA is developed, having taken all different types of users in mind. By using this interface, a common user can pose a question in natural language and get the answer without knowing any of the underlying infrastructure. On the other hand, an expert user can analyze the QA engine and see the output of all the different modules. In addition, users can select different sets of data, over which they want to run the QA engine, as well as different components to complete different tasks. Therefore, we have developed an interface to realize the geospatial question answering engine GeoQA for all users, from the inquisitive scientist to the common layman.
Keywords:
Knowledge Bases, SPARQL, Reactjs, Node.js, Docker
