A pandemic has swept the Chinese New Year holiday into a long and anxious period. The epidemic overlap with the Spring Festival has given people more time to calm down and think deeply. From the perspective of the Internet, after the epidemic, the value of the former “search” returned to sight.
Diverse information acquisition methods have greatly enriched user behavior habits. Under the epidemic situation, users started to search actively from passive information viewing. The real demands of people’s livelihood behind search also make search and public opinion show a kind of “resonance” phenomenon.
In the epidemic, the resonance between search and public opinion
Looking back at the beginning of the epidemic, once again, how the rumors spread rapidly with the help of public panic.
For example, not long after the epidemic occurred, rumors such as “Smoke smokers are less likely to be infected”, “Brine rinses, and vinegar can prevent infection”, “The government arranges the plane to spray medicine” and other rumors are coming. News about the closure of cities, road closures, and restrictions on traffic around the country has also flowed. Recently, there have been rumors such as “taking antihypertensive drugs will accelerate the replication of the new crown virus” and so on.
During the epidemic, the public has higher demands for authoritative information, and the public has therefore given up passively accepting information feeding, but has repeatedly searched to actively verify the authenticity of the information. The number of users actively searching for “new coronavirus” surged. According to data, during the epidemic period, Baidu searched and browsed 1 billion people a day.
In fact, as the epidemic sweeps across the world, under the current epidemic situation, Google search also shows a trend of keyword search related to the epidemic situation. The 10 most popular keywords in Google’s current are coronavirus and lysine disinfectant. (Lysol), Perishable foods, Social distancing, Cancel trip, Solano County, Carnivorous, Dog coronavirus, Beard ( Beards), the movie “Contagion.”
In addition, the epidemic-related words “survival”, carnivores, dog coronavirus, etc., have also increased in the number of Google searches; in terms of travel safety, trip cancellations, safe social distances, safety lines, The contact distance between people has become a topic for people to search. In addition, Google ’s shopping search results page shows products such as hand sanitizers, protective clothing, and masks that claim to be resistant to the new coronavirus. Google said, “Based on advertisers’A combination of price and relevance, such as your current search keywords and your activities. “
This is confirmed to a certain extent. Behind the user’s search behavior, the search data supports the value of public opinion and the insight into changes in public opinion.
In China, TV media also prefers references to search for big data.
CCTV “News 1 + 1” recently cited Baidu search big data for two consecutive issues, reporting on the living conditions of the people in Wuhan in the epidemic. From the data point of view, the number of searches for homemade food by Wuhan residents rose by 1,590 %, The search peak of mustard keywords increased by an average of 1088%.
After major companies have resumed work one after another, CCTV Finance, Beijing Satellite TV, Guangdong Satellite TV, Xi’an Satellite TV, etc. have all focused on reports of companies returning to search for big data.
The core value of the TV report is reflected in the ability to detect public opinion and concerns. Baidu search data has been repeatedly cited by TV stations. It can be said that search big data is closely related to the public’s most direct epidemic prevention needs, changes in public opinion, and user psychology. It also has three characteristics of timeliness, importance and proximity. Because of this, the news value of these search big data is also greatly highlighted.
From the search index of Baidu, we can also see a synchronization and resonance of data trends and public opinion changes.
For example, in the initial stage of an outbreak, game became the first search term. Over time, searches for “masks” and “alcohol” increased, while searches for “psychological counseling” and “throat pain” in epidemic areas soared 74 times. With the start of resumption of school, key indexes such as “remote office”, “online education”, and “return to work” have skyrocketed.
From the TV media report timeline, almost all of them shift the focus of the report around the phased changes in user search keywords. The search keywords and the current public opinion have almost the same resonance phenomenon-every time the heat curve presented by the Baidu search index, it accurately reveals the needs and concerns of the offline public.
It can be said that it is the carrier and barometer of social mentality, social problems, and public opinion dynamics in the epidemic situation.
Why does a search reflect real opinion?
Many internet platforms are paying attention to the epidemic. Why can only search detect the trend of public opinion and public opinion? Of course, the main reason for this is that the search for big data can collect the wishes and behaviors of massive real users with analytical value.
With regard to real user needs, there is a saying, “Customers do not want to buy electric drills, but to buy the hole in the wall.” From the sales story of beer and diapers to the lipstick effect (lipstick sales due to economic depression ), We can all see the power of big data.
This is essentially that the platform has detected its real needs through precise mining of user behavior. However, corresponding to the news information platform, short video platform and social media under the epidemic situation, behind the “crowd of noise” of netizens, there is a lot of useless social information, and the noise and the manipulation of marines and machines are mixed, and information overload becomes the norm. , It does not truly reflect user needs and get real public feedback.
Compared with the noise of platforms such as social media, behind the search is a large number of users actively verifying behavior. What the user is searching for at this time reflects what the user is thinking. These data reflect people’s concerns and can be more accurate, Respond to users’ real wishes in a timely manner. At the same time, Baidu can extract valuable information from the growing mass of data through effective big data processing and analysis methods such as the search + information flow model and Baidu index.
On the other hand, the soil thickness of information and knowledge is different. It is worth mentioning that Baidu will conduct a comprehensive “upgrade of information and knowledge” in 2019.
In the past 15 years, Baidu has accumulated over 1 billion pieces of high-quality content including knowledge content, encyclopedia, and library. Baidu Encyclopedia, Baidu Library, and Baidu know that these three products have established cooperation with more than 7,000 authoritative institutions such as the Chinese Association of Science and Technology, the China Anti-Cancer Association, the People’s Daily Digital Communication, and the National Philosophy and Social Science Documentation Center, andAfter investing in high-quality vertical content such as nutshell, Kai storytelling, and Zhihu, Baidu’s own content system-post it, know, encyclopedia, Baijiahao, iQiyi, and good-looking videos. Docking search.
The thickness of the information determines the richness, comprehensiveness, and authority of the sample, which can better control the cause mechanism and dynamic context of the event. For example, Baidu can provide authoritative knowledge and corresponding solutions for many aspects such as virus interpretation, etiology, medical treatment, treatment, industrial epidemic prevention, and user psychology.
Some research scholars have pointed out that compared with the use of traditional data for benchmark model prediction in the United States, the average absolute error of out-of-sample prediction using search data is smaller.
A case known in the industry is that when the 2009 H1N1 influenza outbreak in the United States, Google completed the prediction of winter flu by observing people ’s search records on the Internet, and the data it measured became the data of the United States public health agency. With very valuable information available, Google has also become a more effective and timely indicator.
Of course, this is largely determined by the thickness, sample richness, and comprehensiveness of Google’s information data.
Search for large numbers to guide people’s livelihood, or topics worth studying
Therefore, how to give full play to the value of search big data for public opinion detection or a topic worthy of our consideration, under the current epidemic situation, its value in guiding people’s livelihood has two points:
One: It can screen public opinion and public opinion trends that are of concern to the whole people, and provide local governments or institutions as a reference for decision-making, and provide feasible observation directions and perspectives for public opinion polls and control of public sentiments.
Second: provide data samples to help more SMEs develop local plans tailored to local conditions.
During the epidemic period, an average of over 1 billion people per day learned about the epidemic through Baidu search, which led to more dimensions of search content and reflected more aspects and details of social and economic issues. For example, Baidu’s search big data report shows that some offline industries have been severely affected by the epidemic, and the tourism industry is the first, followed by the hotel industry, decoration industry and real estate industry.
For example, in Guangdong Province, the industries that have a higher degree of focus are finance, machinery, electronics, electronics, and textiles, which means that these manufacturing sectors have higher demands for resumption of work.
From the case of Google abroad, there has been a review of the practice of Google ’s prediction of flu in the industry. It is to find out the 50 million keywords searched on Google in the United States, and then get each keyword to be searched. A time series. Each keyword will generate a “model”, list the top 100 keywords, enter the next step, and then the method is to use the first ranked word to the model (logit (P) = β0 + β1 × logit (Q ) + ε), get LOGIT (P ‘), then do a correlation analysis with the real LOGIT (P), add the first and second words together as Q, and see the results and the real data the same Correlation, and so on to the first 100 words added up as the result of Q.
If the first 55 words add up to Q, the model has the best simulation degree, and it is closest to the real data. Then the specific independent variable of this prediction model can be determined: the search volume of these 45 words is superimposed as Q That is, if the first 55 words are all related to the flu, and 19 of the next 55 are not related to the flu, then the first 45 keywords can be used to make specific judgments as the qualitative keywords closest to the truth.
This model may not be easy to understand for ordinary people. In short, it is through keywordsJudgment, determine what data we want to predict, determine what we have, and then compare and analyze user search data to distinguish its source, time period, and demand point. A quick and accurate forecast data comes out. If it is for flu predictions, 1-2 weeks faster than official data means more time to prepare for how to respond to a possible flu pandemic.
As mentioned above, the prediction model made through search can also be used for anti-epidemic work after the epidemic. In the follow-up, how should local governments launch relief and subsidies for epidemic prevention and control, material deployment, and related affected industries, and formulate corresponding policies. User search data and behavior trajectories provide a relatively objective and directional reference “road sign” In fact, it can provide local solutions tailored to local conditions. Subsequent anti-epidemic programs that address public needs can also affect public expectations and change public behavior.
Behind the rapid growth of the keyword search related to the epidemic, it reflects a new decision-making idea for people under the Black Swan incident-using traces left by search as data to present the nature of the event and even predict the development trend of the event, It then guides us to better avoid risks and turn crisis into opportunity.
Under the epidemic, the value of searching for big data in reflecting people’s livelihood and serving people’s livelihood has been magnified. In the future, individuals, the media, and government agencies need to further understand the current public opinion and the trends and trends of events. Searching for big data to assist action decisions is a good idea.
As mentioned earlier, search can reflect the staged appearance and demands of people ’s livelihood needs. How to open search big data and connect with local government agencies, academic research institutions, local enterprises and the public, etc. The research on data is very important to promote the solution of people’s livelihood problems. This may also be worth thinking about by technology companies such as Baidu.
And the search for big data in various fields can sort out the logic, background and cause mechanism behind the needs of the people. Through joint action to quickly promote the implementation of targeted solutions, they can quickly respond to people’s livelihood demands and drive the public to participate in society In the management of problems, the same logic and practices can be reused in the field of industry, university, research, and research, which can also improve the ability of various industries to think and solve problems through data.
In general, search is like a mirror that reflects the most urgent needs of the people. Through insight into public opinion, it can provide certain data decision value and output it to local governments, institutions, and the public. To the people, to the people “value.
How can future search better guide people’s livelihood, improve people’s livelihood, or be a subject worthy of study.