The blue ocean of intelligent Voice in private domain scene
Looking at this title, many people wonder, where is the distinction between private and public domains?
"Common domain" mainly means that the current usage scenario is 90% of all users, and the resources and information Shared and required by users are common, and the ways for users to obtain resources are diversified.
"Private domain" means that the user group is unique and partial, and only meets the needs of users in the current scene. The way the information has been obtained is unique, and in a different scenario all of the content will become invalid.
1. Example: Catering industry
All of our users have been to restaurants to order their meals. We can imagine a voice robot helping the store user complete menu recommendation -- assisting the user with basic questions such as price -- to complete the user's checkout behavior.
We will find that the whole process is applicable to all restaurant industries. Robots replace humans to meet users' basic ordering needs. The current scenario is a typical intelligent voice interaction scenario in the public domain.
What is the private domain voice scene?
Based on the positioning of different restaurants and the different products they sell, they provide targeted ordering services.
Can imagine, we go to Chaoshan beef, users will have the above questions "where is the tendon meat?" ", "Is this meat okay for my mother to eat? She has a bad tooth. "" What's your special today?"
You can also imagine, we go to haidilao we will ask what? "Give me a rubber band and tie my hair." "How much do you have for half and one portion?" ", "Please call me when I get my nails done."
Let's imagine that the same problem may not appear in different restaurants at the same time at first. For example, we won't ask the waiter in Chaoshan beef for the two questions of rubber band and manicure. At the same time, we will also find that the two stores give different answers to the same question. For example, today's discount must be different between them.
If two restaurants buy a batch of robots at the same time, after meeting the basic public domain service, basically can't do other things, but also need more human waiters to help customers solve the information problem in the private domain.
Two, for example: car airport view
With the development of voice in smart cars, more and more users will use voice to open navigation, listen to music, check the route and other basic public services in the car.
What do private domain services include?
First of all, in the scenario of car engine, the service of private domain is closely related to the user information. For example, the car prompts the user whether to buy a cup of coffee in a nearby coffee shop, or whether to buy a bunch of flowers on the birthday of their loved one.
Secondly, in the car airport view, and the current car yard or car brand has a close first service; Just think, if the user does not have glass water at present, if he/she asks "where is glass water nearby?", the car and the machine can navigate to the 4S shop in cooperation nearby. If the car does not have this service, the user will open the mobile app to buy glass water, or the mobile phone will check the nearby place to buy glass water. This is equivalent to pushing users from the private domain scenario to the public domain scenario, which is a user's needs are not met.
Of course, there are more information about car QA, car membership service and other private domains that can only be realized under certain scenarios. All information will be different after the brand is changed.
3. Examples: Hotel industry
The hotel intelligent robot can satisfy users' basic check in or other public domain services; However, different hotels due to the different target groups, the positioning of the hotel is different, the service and service content is not the same; When users interact with voice, they get different information and content.
Take the voice service in private domain as an example. For the user group of home stay, mainly travel and leisure, the hotel can voice prompt good places nearby to eat and have fun. Here, we can focus on recommending projects that have cooperation with the current home stay facility, and finally, we can provide services such as delivering specialty products to the home.
For business travelers, business hotels and express hotels are the main choices in the current user group. In terms of voice products, this type of hotel does not need to provide fun and delicious recommendations. Mainly around business and personnel, voice services such as morning call service, coffee service, or cloud video conferencing are needed.
Through the above three examples, it is relatively clear to analyze the different voice services and product positioning of public domain and private domain in the same scenario of intelligent voice dialogue.
If intelligent voice is needed to be more intelligent and become a robot that can communicate, or can meet 80% of customers' needs in a specific scene, it is necessary to do product demand mining and product design in the private domain.
If we still design products for all c-end clients according to the current market conditions, such as xiaodu, Tmall and ding dong, we will not only reach the bottleneck of product development, but also find it difficult to promote the development of intelligent voice interaction again.
Four, I want to talk about the core approach
What should we do in the face of private blue Ocean?
In fact, it only needs to be summarized in one sentence -- according to the specific needs of the private domain, in-depth mining of the private domain requirements and product sorting of the knowledge base.
There are mainly the following three ways:
Data management of demand: Various data and demand mining methods can be adopted to obtain from the interviews of customers and professionals, and to summarize and sort out the problems of users' demands in the current private domain by learning the non-content of words and recording tracking of salesmen.
Data structuring: The collected requirements are processed with structured techniques, and the knowledge contents of different dimensions are extracted according to the content, object and scope of the requirements. A way to move from a single piece of data to a one-to-many, or many-to-one, by making simple connections.
Knowledge-based processing of data content: the structured data sorted in the previous step are handled in the model to complete the network of data content, so as to truly achieve the network of knowledge map; For example, in view of the question whether my mother eats beef, the age of the mother who does not use it, whether she has had dental surgery and other related issues, the knowledge map is formed, and the degree of correlation between the problem and the private domain is extracted, so as to be applicable to multiple private domain scenarios.
The ultimate purpose of intelligent voice is intelligent conversation, which can complete the conversation content of human-like behavior. The development of intelligent voice in private domain will be the next blue ocean.
Private domain voice is also the rigid demand of users' usage and requirements under the private domain scenario.
If the private domain voice product is well done, it will not only help enterprises to reduce a lot of labor costs; At the same time, substantial progress has been made in satisfying thousands of people and providing basic guarantee for the implementation of real intelligent dialogue.