- First challenge is to get relevant and related knowledge from various sources for a topic under consideration.

- Second challenge is to effectively utilize the information gathered to generate meaningful texts.


San Jose State University

“In NLG systems, knowledge is an awareness and understanding of the input text and its surrounding context.”

 Knowledge sources can be grouped into two

 - internal knowledge
 - external knowledge 

Incorporating knowledge into the text generation process is knowledge-enhanced Text Generation.


Knowledge-enhanced Text Generation

A Survey of Knowledge-Enhanced Text Generation

Internal knowledge creation takes place within the input text(s), including but not limited to keyword, topic, linguistic features, and internal graph structure.

Internal Knowledge Source in NLG Systems

External knowledge acquisition occurs when knowledge is provided from outside sources, including but not limited to knowledge base, external knowledge graph, and grounded text.


External Knowledge Source in NLG Systems

Challenges in Knowledge-enhanced Text Generation

Specialized architecture models are designed to incorporate particular knowledge into the text generation models. The various neural network models used to enhance the knowledge  are 

 - Knowledge-related Attention Mechanism
 - Knowledge-related  mode in Copy and Pointing Mechanisms
 - Knowledge-related Memory Network
 - Graph Network
- Pre-trained Language Models


Learn Before

Related