Takako Hashimoto, Director of Chiba University of Commerce Research Center for Economics

The Internet population in Japan is said to exceed 100 million, and more than 80% of the population uses some kind of information device such as a PC, smartphone, or mobile phone[1]. Social media, which is a "two-way communication medium in which an unspecified number of people can participate," has also become widespread, with some surveys indicating that the number of Facebook users in Japan exceeds 24 million[2] and the number of Twitter users exceeds 20 million[3]. On social media, opinions freely posted by people in public tend to be cultivated as a kind of consensus and spread as "word of mouth." In fact, more than 80% of people "collect information from the Internet before purchasing any product or service," and 40% of people "decide to purchase based on word of mouth" when making a purchase.[4] Social media has now become an indispensable source of information for product and brand marketing. Here, I would like to discuss the relationship between social media and marketing.

Word of mouth on social media is thought to have the following characteristics:[4]

(1) Reviews remain and can be referenced later With oral reviews, you can only hear them if you are there, but reviews on social media are usually exchanged as text information, and that information remains on the site. Unless it is deleted, it can be viewed later, which plays a major role in building consensus on social media.
(2) Communication between strangers When communicating through word of mouth, you can recognize who you are speaking to, but in most cases, the people you communicate with on social media are an unspecified number of strangers.
(3) Spreading Word of mouth on social media can sometimes spread at a scale and speed that is beyond imagination. Negative word of mouth in particular tends to spread quickly.

In recent years, there have been many attempts to use word-of-mouth on social media for marketing purposes, such as measuring the effectiveness of product campaigns and brand image research. For example, when measuring the effectiveness of a product campaign, the number of Twitter Tweets containing the product name is measured, and time-series changes and related words (words that co-occur with the product name) are analyzed. By determining whether the related words are words that meet the product developer's expectations, such as "I love it," "It's delicious," and "It was good," the reaction to the campaign is understood and its effectiveness is evaluated. When comparing the image of competing products, comparisons are made from various perspectives, such as the number of Tweets containing the company's product name and Tweets containing other companies' product names, time-series changes, related words, and whether there are more positive or negative reviews. If it is known that a product has a large number of Tweets in the summer, it can be assumed that those products are often used in the summer. Through the related words for each product, it is possible to know the situations in which they are used (e.g., when you are not feeling well, when you wake up, etc.). In addition to general products, it is also possible to compare cities such as Shibuya and Shinjuku, and competitive comparisons of software environments such as Mac and Windows. Through these analyses, it becomes possible to understand the strengths of one's own products, the differences between them and those of other companies, and areas for improvement, and to use this information in marketing. In recent years, commercial services that analyze social media word-of-mouth have been actively developed, and many companies are using these services to obtain the information they need to market their products.

Various information technologies are used to analyze reviews on social media. Reviews are large-scale data, so technology that can analyze big data is required. High-speed search technology is also used, such as instantly counting the number of tweets containing specific keywords. In addition, various data mining technologies such as language component analysis, qualitative evaluation, reputation analysis, text mining, and topic extraction are also applied. Artificial intelligence technology, which has been a hot topic recently, is also used. These various technologies are still in their infancy, and further technological innovation is expected in the future. In that sense, it can be said that analysis of reviews on social media is a field in which the latest research results can be immediately used for business. Reviews on social media are large-scale data that is updated daily in real time, and are a valuable source of information that accumulates people's real opinions. It is expected that the business use of reviews on social media will become increasingly popular. We hope to see further developments in both technical and applied aspects.

References
[1] Ministry of Internal Affairs and Communications, Information and Communications White Paper 2014 Edition
http://www.soumu.go.jp/johotsusintokei/whitepaper/h26.html
[2] Cereja Technology Co., Ltd., Estimated number of Facebook users in Asian countries
http://www.cereja.co.jp/press_release20150323.pdf
[3] eMarketer, Asia-Pacific Grabs Largest Twitter User Share Worldwide
http://www.emarketer.com/Article/Asia-Pacific-Grabs-Largest-Twitter-User-Share-Worldwide/1010905
[4] NTT Resonant Inc., Survey on the impact of word-of-mouth on purchasing behavior
http://research.nttcoms.com/database/data/001436/