Construction of an Intelligent Knowledge-Based System Using Transformer Model

A knowledge-based system (KBS) is a type of artificial intelligence (AI) that seeks to capture the knowledge of human experts in order to aid in decision-making. The process of populating a knowledge base by extracting information from unstructured sources such as pdfs, text, images, and so on is known as knowledge base construction (KBC). It is frequently difficult to deal directly with unstructured data, such as documents. KBC systems assist in storing information in an easy-to-use format. When data is stored in structured format, it becomes easier for applications to use previously inaccessible data. A knowledge-based system’s typical architecture includes a knowledge base and an inference engine. The inference engine derives insights from the knowledge base’s information. Knowledge-based systems also include an interface through which users can query and interact with the system. The main aim is to analyse and extract the key evidence from research papers by constructing a knowledge base using the Transformer Model that helps the teachers identify multiple teaching methods that are effective for the students with learning disorders. The major steps involved in the implementation of the proposed system are text preprocessing and analysis, transformer-based joint entity relation extraction and model evaluation. The dataset consists of full text file (PDF) research papers. The text preprocessing and analysis step defines two pipelines for identifying data and targets. The sentences are then represented as tokens using simple Bag-of-Words and weighted TF-IDF. The tokens are then passed through BERT to extract the relevant entities and predict relations. The joint entities and relations consist of teaching methods as entities. Finally, token embedding is performed to classify the entity.