Information Extraction

This chapter introduces information extraction (IE) as a key technique for transforming unstructured text into structured knowledge, with a focus on named entity recognition (NER) and relation extraction. It first presents a Bi-LSTM-CRF-based approach for NER, detailing data processing, model construction, and evaluation using the CLUENER2020 dataset, where entities are identified through sequence labeling. The chapter then explores relation extraction using ERNIE-UIE, a prompt-based pre-trained model framework that supports low-resource and zero-shot scenarios, demonstrating its ability to extract semantic relations as structured triples. Together, these methods illustrate practical deep learning strategies for advancing automated information extraction tasks.