Sweeping Knowledge Graphs with SPARQL Queries to Palliate Q/A Problems

Question answering over knowledge graphs (QA-KG) seeks to leverage the information stored within a knowledge graph (KG) to respond to questions posed in natural language. This approach assists end users in accessing the extensive and valuable information contained in the KG more efficiently and effortlessly, without requiring them to be familiar with its underlying data structures. QA-KG presents a significant challenge, as extracting the semantic meaning of natural language is a complex task for machines. Simultaneously, numerous techniques for embedding knowledge within knowledge graphs have been introduced. Knowledge Graph-based Question Answering (KGQA) still faces difficult challenges when transforming natural language (NL) into SPARQL queries. Most systems answer simple questions referring only to a triple, but more complex questions requiring complex queries containing subqueries or multiple functions are still a difficult challenge in this research area. This article presents an approach that addresses these challenges by providing several key features. First, it facilitates data annotation to address concerns about misspelled and incomplete metadata. Second, it allows data repair to handle missing values in the dataset. Third, it provides data augmentation capabilities, allowing the dynamic addition of meaningful columns and their corresponding cell values. Finally, he learns to answer questions based on SPARQL queries within the framework of Knowledge Graphs. The effectiveness of this approach has been evaluated using benchmark data sets with promising results.