Conversation model

Get Started. It's Free
or sign up with your email address
Rocket clouds
Conversation model by Mind Map: Conversation model

1. elements

1.1. topic

1.2. context

1.3. type

1.3.1. question

1.3.2. statement

2. Theoritical Problems

2.1. Topic inference

2.1.1. Limit to "film"

2.1.2. LDA?

2.2. Response generation

2.2.1. Simple enlg

2.2.2. openccg

3. Different type of queries

3.1. Question

3.1.1. For questions, we need to find a answer, some times the answer is about the bot or else, so we need ...

3.1.2. Knowledge base

3.1.2.1. This will be a large stuff

3.1.2.2. workarounds

3.1.2.2.1. Just use a chat bot program, which requires no knowledge resource, but still robust in context

3.2. Statement

3.2.1. For statements we needs to make a statement or a question as response

3.2.2. Statements are usually downplayers or expression self opinions

3.2.3. Question is made when confirm or query some knowledges.

3.2.4. Create a new topic

3.2.5. Workarounds

3.2.5.1. Subtopic inference model

3.2.5.1.1. Skip this by ensuring a fixed topic

3.2.5.2. Generate close related sentence without full understanding

3.2.5.3. Decide when to change topic/subtopic, or end the conversation

3.2.5.3.1. It can be done randomly, when user gives a statement

3.3. How to answer them ...

4. Practical problems

4.1. Make a acceptable response given a statement query

4.1.1. Hierachy model

4.1.1.1. Express self opinion on the same / related object

4.1.1.1.1. This usually leads to a coming question

4.1.1.2. Make a confirmation / Query of some informations on the related topic

4.1.1.3. as a last solution: downplayer

4.1.1.4. as a last solution: downplayer

4.2. Topic construction in the response

4.2.1. Input

4.2.1.1. topic, keywords as the context

4.2.2. output

4.2.2.1. Object / topic to respond with