Development Of A Question Answering Model Using The Indobert Algorithm To Evaluate Website Content In Indonesian Language

Technology is currently developing increasingly sophisticated and making the use of digital technology increasingly easier for humans. One example of technological development is the development of natural language processing (NLP) technology to support better information search and extraction functions. This research aims to implement the Natural language processing (NLP) method and the BERT algorithm in the question-answering system to help analyze the relationship between website content and search engine search results. This research uses a literature study method, namely looking for references in the form of books, journals, research report articles, and sites on the internet. Apart from that, in data collection, a dataset search was also carried out regarding the question-answering model needed to create a question-answering system using the IndoBERT algorithm for evaluating website content in Indonesian. Then, the results of this research can be categorized as quite good with an Exact Match (EM) of 60.21% and an F1 Score of 71.36%. This indicates the model’s ability to provide appropriate and relevant answers in a large number of cases. This research was also successful in extracting content from web pages and evaluating it using BERT Score.