In times of increasing political polarization, information tends to be less reliable. To counteract this trend, fact-checks have become more prevalent, verifying relevant claims made by politicians. Traditional fact-checking relies on human research and analysis, which may be less efficient than an automated method. Previous automated fact-checkers such as the SmBoP model attempted to directly translate a natural language query in the form of a question into an SQL (Structured Query Language) query in the form of code using recurrent neural networks pre-trained on a database. However, direct translation can often result in unstructured and syntactically incorrect SQL queries that do not query properly on the database. My proposed research aims to create an automated fact-checker that extracts relevant features from a natural language query and inserts them into a template in order to generate the SQL query, providing more structure to the query. This automated fact-checker would be able to process plain English into an SQL query. In order to extract relevant features from the natural language query, I will use a frame semantic parsing approach to first identify the frame, which describes the general topic, then extract the frame elements, which are relevant ideas under the frame. Given that many claims politicians make are about their voting records, I will focus my research on the voting frame, under which relevant frame elements include the agent, issue, and position. With this approach, I aim to increase the structure of the SQL queries such that they will always be syntactically correct while maintaining the same accuracy as a direct translation method.