Go to Content > Knowledge Training > Review to inspect the articles that have user feedback. A person responsible for managing the articles by checking a query and corresponding results is called an expert.
If Kbot provides a helpful article, an expert should accept the positive feedback from a user (or reject the negative feedback).
If Kbot provides an article that is not useful in this particular case, an expert should accept the negative feedback from a user (or reject the positive feedback).
This approach allows to update the recommendation status, which might be “new”, “accepted”, or “rejected”. This work ensures that articles from external search systems are helpful and analyzed by the Recommender system.
Article details
When you expand an article's pane, you can study general and user statement-specific details. They help to decide whether a particular user statement corresponds to the given article and whether the feedback is correct.
General
Article title;
Search engine that has provided this article;
Language;
The number of different user statements that are associated with this article.
User statement-specific
The text of a user statement;
The number of identical user statements;
User feedback.
Recommendation status
Once a user marks an article as “useful” or “not useful”, the recommendation status of this user statement is set to “new”. Now an expert must decide whether the user statement actually corresponds to the given article and whether the feedback is correct.
An expert accepts the user statement when the article is useful for the given user statement, and the user feedback confirms it. The “accepted“ recommendation status is set.
An expert rejects the user statement when the article is not useful for the given user statement, and the user feedback confirms it. The “rejected“ recommendation status is set.
Adding new user statements
You can enhance the Recommender system by adding user statements manually. To do so, type a user statement in the Add new sentence field and press Enter or click Add sentence. Now this user statement is associated with the article and the “accepted” recommendation status is set be default.
Filters
The Filter pane on the right side of the window allows you to narrow your search. Use it to quickly find a specific article or user statement, to view items with a particular status, etc.
Search by article title: Type a title (or just a part of it) to view matching articles. For instance, type "configuration" to see the list of articles about various configuration matters.
Search by user message: Type a user statement (or just a part of it) to view matching messages. For instance, type "how to" to see the articles sent to users who wanted to learn how to perform a task.
Date from/Date to: View the articles that have been evaluated by users during a specific period.
Search by sentiment: View articles with a specific kind of feedback. You can filter out useful articles and investigate the ones that are marked as "not useful" to improve the situation.
Search by status: View articles with a specific status. For instance, you can filter out accepted articles and focus only on new ones.
Search by knowledge base: View articles from a specific knowledge base.
Search by language: View articles in a specific language.
Add Comment