KAIST College of Business 상세 > > >KAIST COLLEGE OF BUSINESS
본문 바로가기 사이트 메뉴 바로가기 주메뉴 바로가기

Targeting a "Hub" for Marketing ... Reputation Spreads More Quickly by Customer’s Word-of-Mouth2013-01-02Hit:5347

Targeting a `Hub` for Marketing ... Reputation Spreads More Quickly by Customer’s Word-of-Mouth

“The U.S. Center for Disease Control and Prevention (CDC) has created a weekly status report on nationwide flu patients. Flu-patient data from physicians helps doctors figure out where flu is spreading. Based on this report, the CDC decides which areas should be intensively managed. It usually takes about two weeks to collect and analyze statistical data to create a report. The next question, then, is how long does it take for Google to create such a report?”

Professor Jeong, Ha-Woong teaches an Executive MBA class in the AIM program in KAIST Business School. He majored in physics, but he is pioneering a new field called ‘Complex Systems Networks’ based on how Google creates such a report.

"Google published a paper in the journal, Nature, in 2008. Its title is ‘Detecting Influenza Epidemics Using Search Engine Query Data’. Utilizing CDC flu reports published between 2003 and 2007, Google selects 50 key flu-related words such as ‘alleviate fever’. Google assumes that if people search using these keywords in Google, then there must be a risk of flu. Surprisingly, Google’s prediction of flu outbreaks is almost as consistent as the CDC’s flu reports. Moreover, Google predicted flu 2 weeks earlier than the CDC -- ‘real time prediction’

Election results are also predictable

"Google has introduced a very simple technique: laying out web pages ordered by the number of clicks on the web sites. It can be said that Google users are doing what Google should do. In addition, Google finds the web pages that are accessed the most.

Google"s search engine system can also be used to predict elections. This is because typing the person"s name in the search window shows the number of results immediately. "I typed each candidate’s name in Google search box at 23:15 the eve of the Seoul mayoral election. The number of results related to Park, Won-Soon was 5.48 Million and the number of results related to Na, Kyeong-Won was 4.66 Million. The next day, the election result was the following: Mr. Park got 53.4% and Mrs. Na got 46.2%. Using the same method, I could predict that Barack Obama would win the U.S. Presidential election in 2008. Of course, in order to get consistently correct conclusions, it is necessary to calculate the weight of each group such as news items or videos. This is called as ‘Big Data.’ In the myriad of many materials, we might figure out something special.

Identify networks

Why did Google introduce this search method for analyzing links between web pages using computers? It is because Google accurately recognizes the nature of the network, Professor Jeong said. "Highways are a network connecting every city and town, which does not have a ‘hub’. If you want to go to Busan from Seoul through highways, then you have to pass by other cities such as Daejeon or Daegu. However, the airline network is somewhat different. You can go directly from Seoul to Busan and all airports in Korea are directly connected with Seoul, which is a ‘hub’.

According to Professor Jeong’s paper published in the Nature in 1999, the network of web pages is revealed as an airline network, which is a ‘hub’.

Sex networks also have a hub

Professor Jeong said, "The following is an example of a sex network. Even though people do not reveal their sexual relationships, various diseases such as AIDS are transmitted through sexual intercourse; it is important to figure out the sex network.” Fredrik Liljeros, a professor at Stockholm University, analyzed that sex network also has a hub through the survey on 4781 Swedes. His paper was published in Nature, in 2001.

“Fredrik Liljeros could not find who exactly had sex with whom. However, there was a big hub which has more than 1000 sexual partners; there were also small hubs having 100 or 10 sexual partners. In this case, I have to call them Casanovas.“

Professor Jeong explains that networks through e-mail or a phone call are also like an airline network, which has a ‘hub’. The reasons are that ‘the poor get poorer and the rich get richer’ and because of ‘efficiency’. He said, “People think that making friends who have a lot of other friends is helpful for them. For this reason, it is easier for people who have more friends to become a larger hub. In addition, an airline network is more efficient than a highway network since an airline network has a smaller number of processes. In that sense, a simple network evolves into a hub as time goes by.

Establish a new strategy by identifying ‘hub’

Google focuses on the fact that the Internet contains networks with a hub. Likewise, many things can be done by identifying the network. If we know the structure of the gene network, then we can effectively find and kill a hub protein of any virus. We can also efficiently change business structures by identifying networks. For example, HP analyzed e-mail exchanges among employees and reorganized structures to form divisions filled with people who send and receive e-mail frequently. In this case, pyramid structure is transformed to the hub form resembling a spider’s web.

"We can establish a new marketing strategy by identifying hubs in the networks between people. Suppose that we give our prototype products to people for free. If we give two products to each person and say ‘One is for you and the other one is for your friend’. Then it is likely that prototype will become a hub. The person who acts as the hub would use the product and share the experience with others. Without finding what the hub is, we can still set up a marketing strategy using hub.”

http://www.hankyung.com/news/app/newsview.php?aid=2012110135811


Contact : Yu, Eunjin ( ejyu@kaist.ac.kr )

KCB ISSUE