A brief discussion on the significance of social network data analysis

With the rapid development of information technology, more and more people are participating in social networks. People are happy to share their relevant information on the Internet and expand their personal network. Enterprises can even directly influence customers through social platforms. Everything seems to be better because of the emergence of social networks, but there are three characteristics of social networks that we need to pay attention to:

1. Feature 1: False information and real information exist simultaneously on the Internet. This feature is particularly prominent in social networks. People will express their real information on social networks, but they will also send out their own imaginary information, which results in rumors becoming "facts" very easily.

2. Feature 2: Unbounded. Information can be infinitely expanded in the network and even affect reality in the end.

3. Feature 3: Fast. Due to the explosive speed of network information transmission, user information is likely to spread in a short period of time.

Each piece of information released is like the ripples spread by a stone in the water. If no more stones are thrown in, the ripples will gradually fade until they disappear. This is the self-cleaning function of social networks. Due to the existence of the above three characteristics, relying solely on "self-cleaning" is far from enough. If another stone is thrown in at a certain point in the process of ripple diffusion, the original ripples will expand or shrink. As long as the points are found correctly, these ripples may form waves. How to find this information, find these points, expand the positive voice of the brand, reduce and eliminate the negative voice has become the key to the success of enterprises in social marketing. At this time, social network analysis can help enterprises.

Social networks are full of all kinds of information that may become "waves", some targeting a product, some targeting a movie, and some targeting a star - all this information can be obtained for free on the Internet. The value of these comments to enterprises can be said to be huge. If an enterprise has these data and analyzes them, it will be very helpful for improving existing products and the direction of future products. At present, for enterprises, the main focus of social network analysis is to find consumers, analyze consumers, and understand consumers. To achieve the simplest and fastest communication with consumers. This requires finding the circle where consumers are through data analysis, and then finding opinion leaders in the circle. Through opinion leaders, the information that the enterprise wants to convey can be further expanded and radiate the entire circle. In this way, more loyal consumers can be attracted.

There are many interesting research topics in social network analysis. For example, the identification of community circles in social networks, the calculation of personal influence in social networks, the propagation model of information on social networks, the identification of false information and robot accounts, and the prediction of stock market, elections and infectious diseases based on social network information. The analysis and research of social networks is an interdisciplinary subject. In the research process, basic conclusions and principles in sociology, psychology and even medicine are usually used as guidance. The group behavior and future trends in social networks are simulated and predicted through machine learning, graph theory and other algorithms used in the field of artificial intelligence.

The division of social circles is not limited to the relationships that users actively establish, but can also be divided through their implicit circles, such as interest attributes. When two people interact frequently in a social network, are they also real friends offline? From an algorithmic perspective, this is a difficult problem to solve, but if we think about this problem from a different perspective, think about our offline contact information, if A and B have each other's mobile phone numbers, then the possibility that they are real offline friends is very high. Including products such as Fetion, Mi Chat, and WeChat, if we can really make a social network based on mobile phone address books, we can make a comprehensive judgment on social circles through heterogeneous social networks, and its value is immeasurable.

What can companies gain by mining and analyzing social network data?

1. Discovery of potential business opportunities

Through data mining and analysis, we can find out whether a user's activity business district is within the coverage of the company's business district; we can know a user's consumption capacity; we can know a user's preferences and recent purchasing habits; we can know the probability that a user will buy our products; we can know the strategies of competitors.

Zynga, a social game company that parasitizes on social networking sites such as Facebook, has cleverly used user data to mine a lot of business opportunities. According to Bloomberg Businessweek, the social game giant collects about 60 billion data points every day, including how long people usually play games, when they play, what game items they like to buy, etc. The company's math geeks use this data to analyze which people like to visit their friends' farms and cities (games developed by Zynga), which virtual items people like to buy, and how often they give virtual items to their friends. Then they will come to such a major discovery: people who often receive virtual gifts from friends will like to play games more, and those who don't receive or receive them less often will not like to play games. Ken Rudin, vice president of Zynga's data analysis department, said: Based on this discovery, a group of math geeks have come up with a solution - for those players who don't receive gifts so often, we will make it easier for them to find tools to build cities (Zynga games), so that they will not rely too much on gifts from others. Zynga's intention is obvious: analyze user behavior, figure out user psychology, and then provide more targeted services to each user with unique behavioral habits in a timely manner.

2. Crisis warning

Through data mining and analysis, we can monitor in real time some information that may cause crisis to the enterprise that is suddenly released on the network. We can also track its propagation path and find the key nodes. We can use "rocks" to break up its propagation trajectory so that the crisis disappears as soon as possible. When a company faces thousands or even millions of discussions created by netizens on social media, it is impossible to manually judge which word of mouth is beneficial to the brand and which will become a brand crisis. Public opinion monitoring can be carried out around a certain monitoring field or event through a scientifically deployed uninterrupted data collection and analysis process. In the early stage, it is necessary to set up the collection scope and keyword groups.

In the middle stage, the collected data is pre-processed by filtering, grouping, clustering, etc., and in the later stage, the data is analyzed and the brand is informed of its own reputation in the form of analysis reports. We found that Baidu Index showed that there was no search record for "public opinion monitoring" before 2011. However, with the development of social media, brands gradually realized the importance of crisis public relations and also more deeply realized the role of data mining and data analysis in crisis early warning.

3. Effect prediction

Through data mining and analysis, the relevant effects can be predicted in advance by looking at the circles that the company has already controlled, the stickiness of the consumer group, the timing of events, and the investment in communication. This allows the company to spend the least money to get the maximum output. In 2010, HP Labs researchers Sitaram Asur and Bernardo Huberman found that Twitter can be used to understand the changes in people's interests and accurately predict movie box office revenue. They counted the number of times a movie name appeared on Twitter and collected about 3 million tweets related to the movie in 3 months. They found that the frequency of the movie name appeared was strongly correlated with the movie's box office revenue. Bernardo Huberman said: "Our predictions are very accurate." Take the zombie movie "Crazy Maniac" as an example. The research team predicted that the box office of this movie in the United States in its first week of release would be 16.8 million US dollars, and the actual number was 16.06 million US dollars. We analyze tweets and measure the speed at which tweets are generated. And we believe that the faster the tweets about a movie are generated, the more likely people are to watch the movie. "

However, the mining and analysis of social network data are still at a relatively early stage, and the mining methods of large-scale, high-dimensional data are still evolving. At present, many basic problems such as sentiment analysis of text language have not been effectively solved, which has caused some limitations on the in-depth study of social networks. However, with the continuous improvement of the level of artificial intelligence research, especially the research combining cognitive neuroscience with artificial intelligence technology, we have seen new hope for artificial intelligence. When we are truly able to solve these problems, social networks will become a useful tool to help us predict future trends. I believe that by then, companies will be able to use data mining and analysis of social networks to develop a more accurate, extensive and effective social marketing system to better serve the establishment of brand awareness and the improvement of market sales.