Ten years of trees, ten years of wind, ten years of rain and a hundred thousand columns.

? -Teacher Chen

Teacher, an ordinary and great name.

On the road of life, the teacher is either the guide on our study road or the indicator light on our life road. They stood on the three-foot platform with a piece of chalk, tirelessly taught us what they had learned all their lives, ignored our mischief, accepted our youth rebellion and worked silently all the year round.

Today is Teachers' Day. Say to all the teachers first: you have worked hard!

What is the image that comes to mind?

Are those endless compliments?

Hard-working gardeners, preachers, mentors, trees in the classroom, saints, teachers, respected, conscientious, educated for a hundred years, tireless and conscientious. ......

King of Procrastination: When this question is over, we will finish class.

"knife mouth and tofu heart": you are the worst students I have brought up.

"P.E. teacher spokesman": The P.E. teacher has something to do today.

"Most Comfort": It's easy to go to college.

"Eye-catching": Don't think that I can't see anything on the podium.

So how do we label teachers? A flash of inspiration? By the way.

In fact, from a more technical point of view, these labels about teachers are nothing more than the data we have accumulated for a long time (the habits and behavior characteristics of teachers over the years) analyzed (understood and refined by our brains) and combined with our experience and cognition.

In today's listening lesson, we will talk about the little knowledge about labels.

The description language of the teacher image mentioned above is the label. Simple labels, such as "juvenile", "fun" and "excellent", are all labels.

Describing in formal language is to describe the characteristics of business objects in natural language, which is a symbolic representation of the characteristics of business objects.

"Data" expresses a fact, and "label" expresses a conclusion and judgment, which is a characteristic sign obtained by highly refining and summarizing data. It is precisely because of this feature that tags can support users to make decisions and take actions quickly and play an increasingly important role in the era of big data.

"Customer label" and "customer portrait" are words we often hear, so that sometimes we misunderstand "Can we only label customers?" Or "only people can be labeled?" . In fact, everything can be labeled. There are two concepts to be understood here: label subject (business object subject) and label subject object (business object).

1. Label Entity (Business Object Entity)

Things identified by labels can be natural persons, enterprises, equipment, work orders, etc.

2. Label Subject Object (Business Object)

Refers to the examples of things identified by labels, that is, specific subject and object, such as Mr. Wang, Mr. Zhang and Mr. Li.

How many categories can labels be divided into? From different angles, the classification of labels can be divided into many directions. Here, we make the following classification according to the generation mode of the label and the essential characteristics of the label itself.

1, attribute (fact) "label"

A supplement to the attributes of things formed by simple data processing, such as the age, gender and place of residence of teachers.

2. Business labels

Tags extracted, summarized, analyzed and mined according to business requirements can be divided into rule (statistical) tags and model tags according to their generation methods.

√? Rules (Statistics) Tab

Tags based on simple summary statistics of a small amount of data. For example, by counting the number of times teachers often procrastinate, teachers can be labeled as "procrastinators' preferences".

√? Model label

Tags based on a large number of data mining modeling are usually called big data tags. For example, consider the content of the course, the probability of failing the exam in the past, the strictness of the teacher and so on. We can make modeling analysis and label a course as "high risk of failing the course" based on the results of the model.

Why label business objects? What is the function of labels? In a nutshell, there are the following points:

1, scattered information aggregation

Tags are the result of concentrating, processing and refining scattered data information, and the information obtained from tags is more direct and comprehensive. For example, teachers usually like to joke, often mingle with students, and always have a smile on their faces. So many bits and pieces of information, if you use the label "approachable", you can directly summarize the characteristics of teachers.

2. Dominant and recessive characteristics.

Many times, the characteristics of things need to be refined through multiple dimensions of indicators, and complex model analysis is carried out. If there is no mathematical foundation and business experience, the analysis results may be difficult to understand. At this time, if you use labels to summarize, you can effectively lower the threshold of data analysis. For example, through mining analysis, the difficulty coefficient of an elective course is 0.8, the risk index of failing the exam is 0.65, and the value of the course is 0.85. For those who have no analysis and business experience, it is impossible to make a decision whether to take an elective course or not, but it will be easier and more direct to understand if they are directly labeled as "recommended courses that require hard work".

3. Commercial standardization

In a certain business background, tags describe business characteristics in a unified way according to business logic, which reduces the understanding error between different people. After all, 1000 readers have 1000 Hamlets. A score of 98 can be marked as "excellent" on the scale of 100, but it may only be "qualified" on the scale of 150.

4. Data security

As a highly refined data analysis result, tags can effectively avoid the disclosure of detailed information data in the process of data circulation. After all, the label of a "high-income group" will not tell others your real income.

That's all for today's introduction of labels. If you have any questions, please leave us a message.