September 28, 2021

The CCF big data and computing intelligence contest wants to use big data to analyze who steals electricity and wants to do 10 things

The State Grid should always be guilty of stealing electricity because it is trying to find a big data plan to solve this problem.

On September 24, at the launching ceremony of the 4th CCF Big Data and Computational Intelligence Competition, State Grid released two related topics.

State Grid Corporation of China stated that it hopes to carry out anti-tamper monitoring and analysis through scientific analysis of big data in order to increase the efficiency of anti-stealing work and reduce the time and cost of the analysis of electricity theft. At the same time, the State Grid Corporation of China hopes to carry out scientific analysis of power-sensitive customers through big data analysis techniques to accurately identify sensitive customers and quantify the degree of sensitivity, thereby supporting targeted refined customer service strategies.

In addition to the State Grid hope to take this opportunity to solve the above-mentioned problems, Sogou, Ant Financial, AdMaster, Oztech, China Unicom Research Institute and many other companies and organizations also released another nine race questions involving search advertising, O2O marketing, public opinion Analysis, monitoring and identification, computing advertising, driverless, market forecasting, LBS marketing, climate forecasting and other fields.

Lei Feng network (search "Lei Feng network" public concern) learned that the pool of prizes announced in these 11 games has reached 550,000 yuan, and according to the experience of past competitions, not only the majority of student parties, but also enterprises participated in the competition. And social individuals and groups specially participate in the team.

Issue issuers resort to real data

According to the organizers of the Chinese Computer Society (CCF) and the companies that issued the titles, unlike the data provided in the standard database for some of the same competitions, the large amount of real data provided by the competition title may be one of the reasons for attracting a large number of contestants.

Representatives of the State Grid said that the questions they issued were derived from the following aspects:

First, the user's daily power information in a certain area of ​​the power supply area of ​​the national grid will be provided. Second, after the abnormal power consumption information after the service evaluation, these power consumption information are actually abnormal through professional analysis.

After abnormal user information, the contestant can find the characteristics of the user who stole the electricity based on the data, form a portrait of the user, and accurately identify the stolen user. As for the national grid, the monitoring efficiency of stealing electricity is improved, and the loss of electricity is reduced. .

Sogou’s representative said

We provide a 100,000 user-level query behavior, which is 100,000 user search online to obtain real gender, age, education and other data through survey questionnaires in various ways. We hope to take half of the data as training questions and the other half as Test questions to search for behavior to identify a person's age, sex, and education.

China Unicom also said that they provided a detailed record of 30,000 users on the Internet, namely record of online behavior recording and call records for three months, and 10,000 merchant location data and activity data. In addition to the raw data, intermediate process data is also provided on the basis of the raw data.

Of course, the above-mentioned publishers also emphasized that the above information was desensitized.

Or a "training" and "recruitment"

What is the main purpose of this competition?

Lei Fengwang editor noted that at the launching ceremony, experts from the Shanghai Ministry of Education and representatives of some colleges and universities were joining in. It is not difficult to see that many colleges and universities would like to take this opportunity to "train soldiers."

For companies that issue titles, AdMaster representatives bluntly stated that they hope to solve problems in the enterprise through the competition. He said:

Take a case of our company - how to identify false traffic. If this scheme is done well, it can be directly applied to our company's business scene, and it can also be applied to the entire advertising industry. There are several advantages to it. First, for contestants, whether students or organizations, or other competitors, can use various knowledge to directly apply various data to the industry. We don’t just provide bonuses. The bigger one is making a great contribution to the entire industry. Second, if it is a student, you can also give him an offer.

Cheng Xueqi, Secretary-General of the CCF Expert Committee, started the "advertisement" case by case. He told Lei Fengwang editor:

The team of Shanghai Jiaotong University, who won the first prize in the first competition, later won the graduation and went straight to start a business. Of course, this is the policy given by the principal of the Jiaotong University. During the period of starting a business, they can temporarily follow the suspension of study. Now they can get it quickly. The second round of financing.

In the second session, a team took part in the competition of the mass science and technology organizations. At that time, the competition titled “Rapid Discovery of Network Hot Spot Events”. After that method was completed, massive science and technology immediately used this achievement, and later when the Tianjin event occurred. The algorithm used is more precise and timely than all other system services.

Links - what are the problems?

1. Pedestrian recognition in monitoring scenarios

Peculiar identification of pedestrian attributes is an important part of intelligent monitoring technology. The title of this game provides multiple images of pedestrians with annotation information under the surveillance scene. Participants are required to study pedestrian fine-recognition algorithms based on positioning (head, upper body, lower body, feet, hat, bag) to automatically identify pedestrians. Pedestrian attributes in the image. The tagged pedestrian attributes include gender, length of hair, upper and lower body clothing, shoes, bag type and color, and provides an image of the pedestrian's head, upper body, lower body, feet, hat, and package position.

2.Human or Robot?

Compared with traditional TV commercials and outdoor advertising, traffic cheating has always been regarded as a unique drawback of Internet advertising. With the development of network data technology, traffic cheating also presents a trend of scale, machine, and industrialization. Ways include making false traffic through labor, machines, or stealing traffic through technical means, stealing posts. Anti-cheating requires strong data support, including a wealth of data storage, a large number of project accumulation and experience accumulation, and data access in different aspects of advertising. The title requires the contestant to establish a model based on IP, cookie, device ID, access time series, UA information distribution and other behavioral attributes to distinguish between normal user exposure record and cheating behavior record, and mark it.

3. Perspective-based domain sentiment analysis

Sentiment analysis is an indispensable technology in the analysis of Internet public opinion, and perspective-based domain sentiment analysis is a key technology in which sentiment analysis is applied to specific areas. In the emotional analysis of sentences, standing in different perspectives, the sentence's emotional tendencies will be different. This game is intended to be a sentiment analysis task that stands in the data user's perspective for specific sentiment analysis to make data analysis more usable. The topic of this competition can be subdivided into two parts: "view extraction" and "view-based sentiment analysis."

4. Chicken ribs or affordable? O2O coupon usage forecast

The use of coupons to activate old customers or attract new customers into the store is an important O2O marketing method. However, randomly placed coupons cause meaningless interference to most users. For merchants, spamming coupons may reduce brand reputation and make it difficult to estimate marketing costs.

Personalized delivery is an important technology for increasing the coupon verification rate. It can allow consumers with certain preferences to receive real benefits, and at the same time give businesses greater marketing capabilities. The contest provided participants with a wealth of data related to the O2O scene. It is hoped that the contestants will use analytical modeling to accurately predict whether the user will use the corresponding coupon within the specified time.

5. Sogou user image mining in big data precision marketing

In the modern advertisement delivery system, a multi-level user image construction algorithm is one of the basic technologies for realizing accurate advertisement placement. Among them, advertising targeting technology based on demographic attributes is a key technology that is generally applicable to brand display advertising and precision bidding advertising. Population attributes include basic attributes such as gender, age, education, etc. of natural persons.

In the search bidding advertising system, the user obtains relevant information by inputting a specific query word in the search engine. Therefore, the user's historical query has a close relationship with the user's basic attributes and potential needs.

6. Traffic sign recognition in an automatic driving scenario

The detection of traffic signs is a very challenging task. Precise detection plays a decisive role in the subsequent identification and assisted positioning navigation. There are many types of traffic signs, and their size and angle do not depend on them. It is difficult to accurately detect them by themselves. In real driving environment, it is affected by weather, light and other factors, making the detection of traffic signs more difficult. We will provide image data in a completely real world scenario for competition training and testing, so that the developed algorithm can be practically applied in autopilot.

7. Customer abnormal behavior analysis

In order to further improve the safety of social power supply and protect the normal interests of power generation companies, power grid companies, and power customers, the occurrence of theft of electrical activities should be eliminated as much as possible. The topic of this competition is based on data provided by the State Grid Corporation concerning the user's electricity consumption, stop of the electric energy meter, current loss, measurement abnormalities when they turn on the light, electricity stealing behavior, and theft of electricity that was confirmed by the on-site electrician on site. The user list hopes that the contestants will use big data analysis algorithms and technologies to discover the characteristics of the user's behavior, form a portrait of the user's behavior, and accurately identify the user. This will help the system identify the stolen user more quickly and accurately. Electricity monitoring efficiency, reducing the loss of electricity.

8. Customer portrait

After years of development and precipitation, the State Grid has accumulated more than 400 million customer archives data and massive power supply service information on the entire network, as well as company marketing, power grid production, and other data. How to find out about power outage events and power supply stability from massive user data? The incident has a sensitive customer group, and the quantitative ranking of the customer's outage sensitivity has a crucial role for the grid company's customer satisfaction, power service level, improvement of electricity customer stickiness, and route optimization of power supply repair work. .

Participants are required to use the power user's 95598 work order data, power supply repair service data, and power outage information data, and combine the analysis and analysis of the text content of the scientific and technical paper to establish a customer power outage sensitivity model, and to carry out the customer's sensitivity to the power outage event. Quantitative analysis to determine whether the user is sensitive to power outage events.

9. Area of ​​Tibetan Plateau Lakes Based on Multi-source Data

This competition title hopes to construct a prediction model for the lake area of ​​the Qinghai-Tibet Plateau by studying various influencing factors on the change of lake area over the Tibetan Plateau.

For example, by studying the total precipitation, temperature changes, topographic relief, elevation, 2000, 2005, and 2010 lake area and distribution data of the Qinghai-Tibet Plateau from 2000 to 2015, the impact of various factors on lake area changes was analyzed. And forecast the lake area in 2015.

10. Forecast of agricultural product price

Price forecasting is the essence of big data. It analyzes a lot of historical data to predict future price trends and provides decision makers with more powerful data support. In the preliminary round, participants only need to analyze the price historical data provided to forecast the price of the agricultural products that are forecasted for the next fixed time. In the semi-finals, participants are required to use as much data as possible to influence the accuracy of predictions by using other data that are influential to the price. For example: weather data, details will be announced on this website before the start of the semi-finals.

11.Accurate marketing based on user trajectory

Accurate marketing is the new direction of Internet marketing and advertising marketing. How to use existing user portraits to classify users and conduct business recommendations for different categories, especially when users are located at specific locations and businesses, how to conduct business based on users' portraits Match the user and push the corresponding offer and advertisement information through different channels.

Currently, there are two main problem-solving abilities: First, the user's real-time location and the merchant's location are matched, and the second is the match between the user's portrait and the merchant's shop.

The contest title requires the participants to extract the association relationship between the user's tag and the merchant's category according to the merchant's location and classification data and the user's label image data. Then, based on the user's position data within a certain period of time, the user is judged to be within 300 meters of the merchant's status ( For a latitude and longitude of 30 meters in 1 second, the user's location and other preferential information that match the user's portrait are pushed to the user.

Magnetic Scale

Displacement sensor, also known as linear sensor, is a linear device belonging to metal induction. The function of the sensor is to convert various measured physical quantities into electricity. In the production process, the measurement of displacement is generally divided into measuring the physical size and mechanical displacement. According to the different forms of the measured variable, the displacement sensor can be divided into two types: analog and digital. The analog type can be divided into two types: physical property type and structural type. Commonly used displacement sensors are mostly analog structures, including potentiometer-type displacement sensors, inductive displacement sensors, self-aligning machines, capacitive displacement sensors, eddy current displacement sensors, Hall-type displacement sensors, etc. An important advantage of the digital displacement sensor is that it is convenient to send the signal directly into the computer system. This kind of sensor is developing rapidly, and its application is increasingly widespread.

Magnetic Dro Scales,Ditron Magnetic Scales,Magnetic Linear Scale,Electronica Magnetic Scales

Changchun Guangxing Sensing Technology Co.LTD ,