Introduction
The user's vague intentions are gradually focused and dismantled by the AI through conversation, and the user's various operation instructions are usually in the form of interactive cards throughout the conversation flow. Consistency between conversational style and role is also key to the AI experience. In addition, behind every good natural conversation experience between humans and machines, there is actually a set of implicit and systematic experience rules hidden. These are exactly what conversation design needs to define.
WHY conversation design
After clarifying the user’s intention and AI role positioning, you can start designing the conversation process. This process not only involves an in-depth understanding of the user's intention, which helps to design more targeted and effective dialogues based on the user's intention, but also involves an accurate grasp of the AI role. Based on the AI role's personality, the appropriate language style can be selected to build a more realistic and convincing dialogue scene.
In the process of interpersonal communication, both parties in communication often follow a series of implicit or explicit rules in order to achieve specific communication goals. These rules not only cover the semantic level of language, but also include the expression of behavior and intention, thereby effectively advancing the dialogue process and achieving the expected communication purpose.
Similarly, in human-computer interaction, the dialogue system also needs to follow a set of rules to ensure accurate communication of semantics, reasonable display of behaviors, and clear expression of intentions. These rules play a vital role in promoting the realization of user goals and improving the interactive experience. Through carefully designed dialogue rules, the performance of the human-computer dialogue system can be optimized to make it closer to the fluency and efficiency of natural language communication, thereby providing users with a better interactive experience.
WHAT is conversation design
Dialogue interaction components are the core content of building a human-computer dialogue system. They generate responses based on preset rules based on questions raised by users. These components are designed to accurately convey semantics, behavior, and intent to facilitate the realization of user goals, and constitute the basic unit of conversational interaction.
**Use different dialogue interaction components to form diverse dialogue expressions: **
- By selecting and combining different dialogue interaction components, a variety of dialogue expression strategies can be constructed. These strategies can not only cover a wide range of language styles and contexts, but also adapt to the individual needs and preferences of users.
For the same user request, different dialogue design components can be combined to form dialogues with the same semantics and different styles:
- Semantic consistency and style diversity: Even in the face of the same user request, by flexibly using different dialogue design components, you can create dialogue responses that are consistent in semantics but unique in style. This design allows AI to provide diverse user experiences while maintaining the accuracy of information delivery.
Therefore, the design and application of dialogue interaction components are the key to achieving efficient and personalized human-computer dialogue. By carefully building and optimizing these components, the performance of the dialogue system and the user's interactive experience can be significantly improved.
HOW principles
In human-computer dialogue interaction, although dialogue interaction components have the advantages of naturalness and simplified operation paths, they also face some challenges. For example, inaccuracies in intent recognition may lead to incorrect replies, and the lack of slot information may require multiple rounds of dialogue to complete the information. The core goal of conversational interaction is to solve user problems and improve efficiency, and any conversational interaction design should follow this principle. In response to these problems, we have formulated universal principles for dialogue interaction, aiming to optimize dialogue design, give full play to its advantages while avoiding disadvantages, and solve users' actual problems more effectively. Through the following principles, a more efficient, accurate and user-friendly dialogue interaction system can be built.
1. Information should be complete and truthful
In human-computer dialogue interaction, it is necessary to ensure that real information is provided to users, establish trust between users and AI, and emphasize user-centeredness.
1. The AI must provide truthful information
The conversational interaction system must provide information based on facts and data to ensure that users can make decisions based on true and accurate information. This requires AI to indicate the referenced data source when processing user requests and ensure the timeliness and accuracy of the information.
2. The AI must communicate its limits
AI should clearly inform users of its functions and limitations to prevent users from misunderstanding AI capabilities. This includes honestly explaining the situation to the user and providing alternatives or suggestions when the AI is unable to provide a definite answer or perform a specific task.
3. Provide targeted information
The dialogue subjects should be targeted and provide specialized support for specific decision-making issues and decision-makers. This means that AI needs to be able to provide customized information and recommendations based on the specific needs of the user to enhance the relevance and effectiveness of decision-making
2. Wording should be clear and easy to understand
In dialogue design, the wording used should be easy for users to remember, understand, and clearly express meaning, thereby achieving more effective communication.
1. Task relevance
The dialogue content should closely revolve around the user's tasks and goals, ensuring the relevance of information to increase the user's attention and retention.
2. Everyday vocabulary
Choosing commonly familiar and easily understood vocabulary helps reduce the user's cognitive load, making information transmission more efficient.
3. Consistent terminology
During the dialogue process, the use of specific terms or concepts should remain consistent to avoid user confusion.
4. Concise sentences
Avoid complex sentence structures and use concise, direct expressions so users can quickly grasp the key points.
5. Clarity
Information must be presented clearly and unambiguously to ensure users can accurately understand the content.
3. Be natural, friendly, and respectful
In interactions between AI and users, it is necessary to respect users and recognize their feelings.
1. Natural conversation
A natural conversational style should be adopted to make the dialogue closer to daily communication, improve the user's communication experience, and ensure a friendly and easy-to-understand conversation.
2. Respect and acknowledgment
In all interactions, the AI will always maintain respect for the user, acknowledge and value the user's feelings and opinions, in order to build trust and a positive interaction environment.
3. Avoid sensitive topics
Sensitive topics that may cause controversy or discomfort should be avoided to prevent unnecessary misunderstandings or conflicts.
4. Handle content carefully
For information or content that the user has not actively requested, caution should be exercised to avoid excessive interference or providing inappropriate information.
Chat component preview
To more intuitively understand and use dialogue components, we divide the dialogue process into four main stages, each corresponding to specific components. In order, they are: Welcome Component, Guidance Component, Follow-up Component, Prompt Component, Confirmation Component, Error Handling Component, and End Component. For detailed information about these modules, please refer to the following document.
| Conversation Flow | Conversation Component | Details |
|---|---|---|
| Awakening | Start | Link |
| Recognition | Follow-up | Link |
| Hint | Link | |
| Confirmation | Confirm | Link |
| Error | Link | |
| Feedback | End | Link |
Original React text: Introduction.