This one is about knowledge extraction. It arranges and supplements two notes NLP notes information extraction and NLP notes relation extraction in school, and combs the basic methods of knowledge extraction, including traditional machine learning and classic deep learning methods.

The "knowledge" involved in knowledge extraction is usually clear and factual information, which comes from different sources and structures, while the methods of knowledge extraction for different data sources are different, so it is useful to acquire knowledge from structured data D2R, the difficulty lies in the processing of complex table data, including nested table, multi column, foreign key Association, etc., the difficulty lies in data alignment, knowledge wrapper from semi-structured data, and wrapper In this paper, we mainly talk about how to acquire knowledge from text, that is, information extraction in a broad sense.

Three most important / concerned sub tasks of information extraction:

• Entity extraction is named entity recognition, including entity detection (find) and classification (classify)

Entity extraction is named entity recognition, including entity detection (find) and classification (classify)

• Relation extraction is usually called triple extraction. A predicate has two arguments, such as founding location (IBM, New York)

Relation extraction is usually called triple extraction. A predicate has two arguments, such as founding location (IBM, New York)

• Event extraction is equivalent to the extraction of multiple relationships

Event extraction is equivalent to the extraction of multiple relationships

This article is mainly about entity extraction and relationship extraction. Next, we will go to event extraction. It's too long. Here's a directory. If you are interested, please directly stamp the original~~