Besides PDF, URL, Documents and Question-Answer-Pairs, CSV-Files can be added as resources to the knowledge base.
A CSV file describes a table that has a fixed structure. The title line describes the information/data contained in the individual columns. It is mandatory and it is recommended to name the columns as precisely as possible.
The title of a column must be unique to enable correct assignments. 
When importing the CSV file, it is possible to define which column titles are to be imported. By default, the entire table is imported. By deselecting the unimportant titles, parts of the table are excluded.
The title columns describe the part that users are most likely to ask about. The columns to be imported describe the data columns that the AI accesses in order to answer queries.
The greater the number of imported columns, the greater the potential for inaccurate or incorrect answers.
It is recommended to use CSV files in specific topics and not just in the general knowledge base. Experience shows that the potential for errors is lower when included in specific intents, recognition is increased and the general quality of the answers is higher.