Datasets:
baoy
/
multilingual_si_testset

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
1
353
"Next Monday."
Then the bot asks, "When would you like to check out?"
The user becomes unhappy here.
I've already said I want to stay for three days from next Monday, why ask again?
Shouldn't you be asking me where I want to stay?
Why is this happening?
It is related to the slot-filling logic of UNIT.
When some configurations are unsatisfactory, some dialogues as weird as that on the left will result.
Let's take a look at the slot-filling logic of UNIT first.
After UNIT identifies the intents and slots naturally, what will it do?
It will check the schema of the intents and slots to see whether the trigger conditions are met or not.
If not met, it will make some clarifications as in the dialogue on the left.
But the clarifications are made in a certain order.
What is the order?
The clarifications are made in the order that the slots were defined .
When the configuration is unsatisfactory, for example, the order of defining slots becomes check-in date, check-out date, number of nights, and hotel name.
If the check-out date is missing here, UNIT will clarify the check-out date instead of the hotel name first, which can be troublesome.
To avoid such unsatisfactory configurations, there are two solutions.
The first is to reorder the slots, putting the hotel name at the very front so that it will be clarified first.
This can solve the problem.
The second option is to split the intent, that is, to split the intent of reserving a room into two.
One is to reserve a room based on number of nights, and the other one is to reserve a room based on the check-out date.
This may be a better solution.
For example, when reserving a room based on number of nights, the slots are hotel name, check-in date, and number of nights; when reserving a room based on the check-out date, the slots are hotel name, check-in date, and check-out date.
In this way, the weird question in the above dialogue can be avoided.
OK, let's take a look at the third case on how to deal with if-else dialogue logic.
The dialogue in the third case is about increasing credit limit.
When the user requests to increase their credit limit, it is necessary to assess their qualifications.
This is the third task of dialogue management as we mentioned just now, and that is to trigger dialogue logics in specific situations.
The bot judges the users to be qualified.
Then, what will it say?
"OK, your credit limit can be increased up to 50,000 yuan. What amount would you like to increase it to?"
This is the general logic.
To establish if-else dialogue logics, I suggest using the DM Kit introduced below.
DM Kit is a local dialogue management module which can run within the developer's own machine environment and integrate seamlessly with our UNIT cloud.
Because it is a local module, it allows you to write codes to perform some complex multi-state processes.
In addition, it also supports databases and Web Hook mechanisms, and can access some external resources.
For example, in the above case, the information that the user can increase their credit card limit to 50,000 yuan must have been acquired by querying the business base.
DM Kit can provide good support for this.
Finally, since it is an open source module, we can flexibly develop and extend it.
To find out more about DM Kit, you can visit the GitHub website through the link below.
OK.
That's all for this lesson.
Thanks！
Developers, entrepreneurs and friends, good afternoon!
It is an honor for me to welcome all of you to attend the ceremony!
Today is the last day of our 2017 Baidu AI Developer Camp, and the openning day of our Baidu AI Accelerator.
We all know that AI is now very popular, but what does AI mean to us actually?
All of us know this saying, "Science and technology constitute a primary productive force."
Over the past 200 years, mankind has already gone through three industrial revolutions, each of which has greatly increased the productivity through injecting the power of technology into every aspect of our production and daily life.
Now we are undergoing the fourth industrial revolution.
So what is the key technology this time?
Artificial intelligence, of course.
AI will not only improve the efficiency of existing industries, but also bring about a lot of new products and new industries.
In this process, although this time the booming of AI is largely due to the Internet industry that possesses an enormous amount of data, very strong computing capability and computational platforms, and many advanced algorithms.
But the impact of AI is far beyond the Internet.
It will definitely affect all kinds of industries and the life of people in all aspects, bringing productivity increase.
And productivity increase, in return, will eventually affect the production relations.
Most of us might have learned this when we were in school.
Economic base will eventually determine our superstructure.
In other words, along with its growth, little by little, AI will bring profound impacts on everyone of us and on the whole society.
Baidu started search engine service nearly 18 years ago.
At the beginning, we would say search engine is naturally an AI product, so from the very beginning of our search engine development, Baidu has been working on some AI technologies, for example, NLP technologies.
And seven or eight years ago, Baidu started setting a full layout for AI.
For example, NLP, speech, image, machine learning, knowledge graph, big data and etc. were started in Baidu at that time.
Now we have got a rather complete layout, from the most basic layer to the perception layer, which covers speech, image, vision, video, AR, VR，to the cognition level, including NLP, KG and user portrait.
More importantly, based on these technologies, we have established a complete platform and ecosystem, which are closely related to every developer and entrepreneur here.
As the host mentioned just now, since October our Baidu AI Developer Camp has been held in many cities.
Today it returns to Beijing and comes to an end.
Of course, its end in this year does not mean it is over.
I believe that more and more developers will achieve their success in our camp next year.
In addition to the Camp, in addition to the AI capacities we offered to AI-related developers and entrepreneurs, we also hope we could provide comprehensive support, in terms of partnerships, market channels and funds.
That's why we have this AI Accelerator that opens today.
We expect that our Practice Camp can help the participants to improve their business skills through practices and exercises, and we wish that our AI Accelerator could provide a much more comprehensive support for all of you.
Before I left the company for the conference, I got an e-mail.
Many of you may know that we held the 2017 Baidu World Conference on November 16th, opening many new capabilities such as AR.
Now, half a month has passed.
The e-mail forwarded by a colleague of mine, said that after the opening of the capabilities, a developer soon has published his self-developed product based on these capabilities.
During the period from November 16th to today, the product has been sold in market and won orders.
And this is just one example I've heard.
I believe that some of you here may have similar experience.
I hope that Baidu can provide more support for you and we will succeed together!
The success of all developers and partners is the success of Baidu AI, and also the success of Baidu.
Let us contribute our share to achieve a better future for our society and our life!
Thank you!
Hello, everyone!
Our course is about the art of coding.
To begin with, I'll introduce myself.
I'm the technical leader of the BFE Team of the Operation and Maintenance Department.
I'm also a member of Baidu's Python and Golang code master comittees.
If you want to know more about me, you can visit my wiki page.
Firstly, I'd like to talk about why we have this course.
Many of you have just graduated from college and come to Baidu.
I'd like to point out first that programming in Baidu is quite different from that in college.
When you were students, you programmed for your course assignments or your teachers' projects.
So I assume mediocre coding would just do.
However, in Baidu we are making industrial products for hundreds of millions of users, which requires codes with higher quality.
Secondly, I want to clear up some of your misunderstandings through this course.
Many of you call programmers “coding farmers” in a self-mocking way.
Many of you believe that you'll no longer have the energy to continue writing codes after the age of 35.