Browse Prior Art Database

Dialog Order Adjustment Method in a Dialog System

IP.com Disclosure Number: IPCOM000249528D
Publication Date: 2017-Mar-02
Document File: 7 page(s) / 91K

Publishing Venue

The IP.com Prior Art Database

Abstract

This disclosure discloses a dialog order adjustment method to optimize the dialog order automatically based on user’s interaction with dialog system. Meanwhile, it also provide a method to predictively optimize the dialog order of seldomly used categories. It can improve the efficiency of dialog flow, capture user need with less questions, reduce the work load of expert as there is no manually operation while adjusting the dialog order. It can also design personalized question order for each user and each user group.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 36% of the total text.

1

Dialog Order Adjustment Method in a Dialog System

1. Background

1) A dialog system will always ask you a series of questions so as to capture your true needs. Such approach will very likely to provide you what you really want. 2) For those dialog systems that use the dialog service, they usually start by asking for what kind of service or product that user wants. Then based on the user’s answers,

the system responds with a series of questions following a pre-designed dialog flow for that service or product. 3) Currently, such dialog flows are mainly designed by expert, based on his or her knowledge of the importance of each questions for each product category or service

category, then sort the question order for each dialog flow accordingly. 4) The expert defined dialog order, is not probably the most efficient way to capture what users really want. It is to say, the dialog order that expert defined, can always be

improved to capture user need with less questions. 5) When the number of product categories or service categories in a dialog system gets large, the amount of manual work for the expert to define the dialog order would

become huge and eventually, impractical for humans to handle. 6) There is always a factor of human errors, either caused by carelessness, subjective, or inexperience. 7) It’s very difficult for expert to define personalized dialog order for all the specific person or specific group of person .

2. Proposed Solution

We propose a dialog order adjustment method to optimize the dialog order automatically based on user’s interaction with dialog system. Meanwhile, we also provide a method to predictively optimize the dialog order of seldomly used categories.

2

3. Benefit

1) Improve the efficiency of dialog flow, capture user need with less questions. 2) Reduce the work load of expert, there is no manually operation while adjusting the dialog order. 3) Design personalized question order for each user and each user group. 4) Optimize question order to improve overall dialog efficiency.

4. Business value

1) Optimize dialog order for dialog system so as to improve dialog efficiency. Normally, the dialog order is designed manually by expert. It will need to be updated again if the requirement changes, what’s worse, it will always bring into some mistakes by manual operation. However, with this proposed method, a dialog system will be automatically updated to reach the most efficiency way.

2) Predictively optimized dialog order for seldomly used categories. For frequently used categories, we can optimize them by analyzing history logs. However, for the seldomly used categories, it is always optimized manually before, which requires large human work. With this proposed method, a dialog system can predictively optimize the dialog order of these seldomly used categories.

3) Help expert to review and update question order. For those questions that are sort to the end of the dialog flow, expert should consider if these questions are...