Mining Relationship Value in Social Big Data 【Dr. Cheong's Data Insights 7】

Newspaper Clippings and Social Media: Understanding Their Importance

Newspaper clippings have historically been an important means for businesses or government agencies to grasp the effectiveness of brand promotion, activities, or policy implementation. This is mainly done through manual clippings or traditional electronic databases.

Social Media as the Main Platform for Public Sentiment

About ten years ago, with the optimization of search engine functions and the emergence of social media, many professionals in marketing, public relations, and media relations began using common search engines or vertical social listening tools (also known as social listening analytics) to enrich and supplement the scope and diversity of clippings. However, their application was not as popular as clippings. Today, social media has become widely popular, becoming the main channel for netizens to obtain information and express public opinion. If decision-makers do not pay attention to social media listening, monitoring, analysis, and mining, they will lose the opportunity to grasp the latest information, gain user insights, anticipate trends, and make effective decisions.

The definition of social media can be very broad; any platform on the internet where netizens can publicly publish and exchange content can be called social media. Examples include social networks like Facebook, LinkedIn, and Twitter; media sharing sites like YouTube and Instagram; social news platforms like Reddit; collaborative knowledge platforms like Wikipedia and Zhihu; and review sites like TripAdvisor and Dianping. In recent years, there have also been many news and commentary channels, such as the Huffington Post and Toutiao.

The types of social media listed above constitute a vast source of social big data. Collecting this information systematically is far beyond what traditional clipping methods can handle. Even using computers and traditional database technology may not achieve real-time collection and classification, let alone analyze and mine valuable parts. Since social big data does not have the clear language norms, content structure, format, and presentation regularity of newspaper content, it requires the use of new big data technologies and algorithms built into social big data mining tools (Social Big Data Analytics) and a team with specialized knowledge to handle it.

Clippings = Monitoring? In-Depth Mining Needed

Traditional information collection, clippings, or social listening analysis tools produce information-level outputs. They can only count quantities and provide general classification results, but they cannot perform the listening, monitoring, analysis, and mining functions of social big data mining tools. These tools can conduct real-time in-depth analysis of the content posted by netizens on social media and the relationships between netizens, using content-based mining and structure-based mining techniques.

In the previous article, I introduced text mining technology. Further understanding of this is what is referred to here as content mining, which interprets the semantic level of the content published by netizens, such as preferences for a certain brand or service, support or opposition to a policy, reasons for high or low consumer satisfaction, etc. Structure mining technology, also known as social network analytics, refers to mining the relationships and interactions between entities or references (such as netizens, institutions, products, events, etc.) in social media. For example, the relationship between a Facebook page and its fans, the interaction between the original poster and respondents on Twitter, and the relationships between keywords (which can be names of people, institutions, or products) in numerous posts.

These mining results are usually expressed through visualized network graphs, which include nodes representing connected entities or references and edges representing their relationships. In market or policy planning and decision-making, if the connections between entities can be displayed through charts and their relationship values can be mined, it can help in the rational allocation and investment of resources. This can form and optimize channels for recommending products and services, enhance brand awareness, and promote and implement favorable policies.

Dr. Angus Cheong Chairman of the Asia-Pacific Internet Research Alliance and Chief Data Consultant of uMax Data

(Originally published in Hong Kong Economic Times, reprinted with permission)