Use of AI knowledge map in natural language processing

As knowledge maps replace traditional databases, the former has increasing wide applications. Here is a case study of how iQiYi uses a knowledge map in its customer service product.

1. Data collection and pre-processing Using technologies like named entity recognition or entity resolution, a database is built. In the example of users’ queries about recharging, relevant data include the customer hotline, types of information required for recharging, etc.
2. Constructing a knowledge map The knowledge map can be built by identifying entities, attributes, and relationships between them. Details can be seen in the pictures attached.
3. Storing the knowledge map The knowledge map can be stored in a graph database like Neo4j, or OrientDB if the amount of data is huge.

When iQiYi’s chatbot receives queries from users, it will retrieve relevant entities and attributes from the knowledge map before returning answers.