Template Discovery for Neural Question Answering over DBpedia

---

Project Description

Nowadays, data is increasing at a rapid rate and is being structured efficiently by the Linked Data Cloud. The data is stored in a specific format and can be queried using the SPARQL language hence it becomes difficult for lay users that are not familiar with formal query language such as SPARQL to search their interests. Question Answering (QA) systems have worked very well to solve this problem. The Neural QA model focus on seq2seq learning to translate natural language questions to their respective SPARQL queries. The goal of the project is to make our end-to-end system perform better via the automatic generation of natural language question (NLQ) templates, the inclusion of new varieties of questions that have advanced SPARQL queries, improving the compositionality of questions, and enhancing the existing generator and learner module.

Implementation Details

The basic ideology, implementation, evaluation, project timeline, and rest of the information can be found in detail in my proposal here.

Mentors

Tommaso Soru, Anand Panchbhai, Nausheen Fatma, Sanju Tiwari