Browse Prior Art Database

Learning the users knowledge of subject matter through discourse in a dialogue system.

IP.com Disclosure Number: IPCOM000246139D
Publication Date: 2016-May-11
Document File: 3 page(s) / 106K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system that focuses on a Dialogue Service analysing the user's language to determine how best to shape its responses. The dialogue service will communicate with the user via common technical language the user is are comfortable with/is appropriate for the context. This allows a rapport to develop between the user and Dialogue Service, lowers the level of confusion and sense of feeling patronised by the user.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 43% of the total text.

Page 01 of 3

Learning the users knowledge of subject matter through discourse in a dialogue system.

Today, there are a variety of different dialoguing systems on the market. They utilize the "ask-respond" framework, and some store details like the person's name, age, birth date. Some store more abstract concepts, such as mood and tone as the conversation evolves; however, the user's knowledge of the domain is not taken into account when the system later shapes responses from dialogue. Simple versions of this more r responsive algorithm do exist where the user has given details, but not during the normal discourse.

    Although most Dialogue Services will use natural language processing to understand the context of a conversation, they respond with the language of how they have been programmed. This is unlike how you talk to a person, where the

response depends on if the dialogue can be perceived as patronising or in a language that the user may not understand, or just have further questions. Existing dialogue systems also use a method called "Conversational Repair" to attempt appropriately respond. This occurs when the Dialogue Service is not fully understanding the context of what is being asked, and will respond with a prompt to drive the user to a more focused response. The most simple form of this is the "Did you mean .... ?" repair response.

    While such systems work in normal human dialogue, in a computer setting, the user is more inclined to get impatient. One method of repair used is to be more explicit. But this then causes the conversation to be more impersonal. This leads to a more explicit method to repair the dialogue, which can come across as impersonal

or robotic. Another method of conversational repair is "grounding" or "being on the same page". This disclosure does not attempt to address this functionality of conversational repair, although its use could enhance the grounding response of the dialogue service.

    Instead we focus on the Dialogue Service analysing the user's language to determine how best to shape its responses. The dialogue service will communicate

with the user via common technical language the user is are comfortable with/is appropriate for the context. This allows a rapport to develop between the user and Dialogue Service, lowers the level of confusion and sense of feeling patronising by the user.

    Additionally, this allows the Dialogue Service to decrease the number of times it is required to do a conversation repair, and instead shape grounding repairs

which have more a more "human" flow to them.

    This will be achieved by using sentence parsing, the system is able to determine the knowledge level of the user, as well as their domain knowledge about

which the dialogue service is talking. Once the level is detected, the system then can offer alternate responses, shape existing responses, or have a representative get in contact to assist with users directly.

    Having systems that adapt to different human knowledge levels...