Answer Type Prediction: Unveiling the Power of Deep Learning Models

Answer type prediction plays a crucial role in natural language question answering systems, enabling the generation of relevant SPARQL queries, a query language for retrieving data from RDF databases. This research paper focuses on answer type prediction for SPARQL query generation. Answer type prediction involves identifying the expected type of the answer to a given question, such as a person's name, a location, or a numerical value. By accurately predicting the answer type, the subsequent SPARQL query generation process can be tailored to retrieve the desired information from a knowledge base. The proposed work evaluates the performance of four deep learning models, which include GRU (Gated Recurrent Unit), Bi-GRU, LSTM (Long Short-Term Memory), and Bi-LSTM. The study's conclusion highlights GRU as the top-performing model for predicting the answer type based on the analysis of input natural language queries. The performance of the answer type prediction model is evaluated on a publicly available dataset, demonstrating its effectiveness in achieving an accuracy of 81% using GRU. The results obtained from the study emphasize the importance of accurate answer type prediction and provide promising outcomes.