-
A product based on Deployment help.
-
Is the writing/improvement of regular expression questions done manually by Nota’s staff?
-
-
Difference from chatbots
-
“Chatbots” that can answer questions in natural language are currently being anticipated. Chatbots utilize advanced natural language processing and machine learning techniques, which is technically intriguing. However, there are many challenges in using them for help systems.
-
① Requires a large amount of data for machine learning
-
It is difficult to train without a system that already has a lot of question and answer records. Otherwise, you would need to generate a large amount of sample data.
-
② Difficult to understand the reason if it doesn’t work well
-
Since the learning system and results are black boxes to the user, it is not possible to perform detailed tuning.
-
③ Tiresome to ask questions in natural language
-
It is difficult to input precise sentences. Can you ask questions in Osaka dialect?
-
④ Only returns one answer at a time
-
Even if there are multiple candidates, only one answer is provided. However, the correct answer may be around the third candidate.
-
Chatbots may seem convenient and fun at first glance, but it seems difficult to use them effectively over the long term.
- I see (blu3mo)
-