Overall reflection about my course participation

Learning process

When I look back to the whole semester learning, feeling like getting into a brand new world. The overall learning process could be roughly concluded in Fig.1. When we start this course, we firstly should have some senses, including the basic concepts and definitions. At the meantime, blogging and commenting to other classmates could not only enhance our understanding to the concepts and open our horizon, but also bond us tightly and made the data we need later. After grabbing enough basic knowledge, we came into the two major parts of this journey: sentiment analysis and social networks analysis. With the help of programming tasks, we get more familiar with theses analysis ways in real life.

Fig.1. Course structure

Online interactions

After visiting and commenting each other’s blogs, the programme assignment 1 was built on the sentiment analysis of received comments. Based on the provide positive and negative dictionaries, dealing with the cleaned data, a sentiment score turned out.

Fig.2. Sentiment analysis formula[1]

According to the above function, my final sentiment score is 0.03328290468986384, which is a little bit larger than 0, suggesting the sentiment of received comments is prone to be neutral. With the respect to the collected negative and positive words in dictionaries, however, the final score may have some bias. The negative words were nearly not about their attitudes towards to the blog itself, but their reviews of the social reality in blogs. In the comments, sometimes we not only talk about the blog content or structure, but also share other different aspects or truth.

Fig.3. My sentiment words

In the assignment 2, according to the social matrix that TA provided, undirect graph was made and we could get four important indexes: Degree Centrality (IN & OUT), Betweenness Centrality, Closeness Centrality. After running the programme, my results are showed in the following:

Closeness Centrality0.501638
Betweenness Centrality0.034397
Fig.4. My network indexes

The high In-degree means there are several different classmates coming to visit my blogs and leave their comments, suggesting the content of my blogs could be, to some extent, interesting and attractive. Out-degree as 13 means I’ve commented 13 different classmates, implying I could probably be an outgoing or social girl. The closeness centrality as 0.5 could a nice result, because it could slightly prove the distance of me and other nodes are not quite long, in other words, I’m close to the other classmates. Besides, the betweenness centrality measures how important a node is to the shortest paths through the network [2] and I’m not sure my result is the best.

Fig.5. Social networks of our class [3]

During this one-semester learning experience, I’ve met and made some great friends, even though we might could only talk through the Internet. I will always treasure this precious time and love u all. Hope u could remember me.❤


[1] IEMS5720 Social Networking Lecture 7,CUHK, p55,2020.

[2] Betweenness centrality. Retrieved from https://www.sciencedirect.com/topics/computer-science/betweenness-centrality

[3] IEMS5720 Social Networking Lecture 12,CUHK, p48,2020.

How SMA can be applied to improve existing health problems?

In blog 1, the definition and process of social media analytics are explained; In blog 2, the primary contents of SDG3 are illustrated. Continue to the last two blogs, how can we combine these two together? In other words, how can social media analytics help to solve the existing health problems? Instead of giving a broad introduction, I would love to discuss some particular issues that SMA can contribute to.

1. Improvement on health insurance

In 2018, I’ve known a “drug hero” named Lu Yong through the movie “Dying to survive” and noticed a group of people are not only suffering from serious diseases but also taking a heavy economic burden. In 2003, Lu was a leukemia patient and spent out all his savings because of a curing medicine Gleevec, which is not covered by health insurance and would cost 23000 yuan just for one-month supply. Leukemia patients like Lu started to share their miserable experience and complain the high cost of Gleevec on BBS. Patients were trying different solutions and seeking help through various approaches, however, without money, they could do nothing but die. Luckily, Lu found an Indian generic version of Gleevec that only cost 200 yuan and started to help patients purchase this drug on BBS and QQ, saving hundreds of lives.

It’s a moving story, but never is an eternal solution for a country. As shown in Fig1, in 2014 China, about 30% of people has to spend their own expense on health matters. Unlike 2003, recently, people love to share and post their lives on social media like BBS, Weibo or Twitter. To prevent this tragedy happening again, SMA can play an important role to help improve the health insurance coverage through the following aspects:

  1. According collected statistics on social media, it’s much easier to detect the existing medical problems (such as the searching numbers, topics, trends).
  2. Help letter on social media can be another important insight. Applying SMA, summarization of existing issues can be easily shown, and the relative organization or government can offer help depending on the emergency. (e.g. Financial difficulties of leukemia patients would be discovered faster)
  3. According to different types of diseases, using keywords like ‘money’, ‘specific drug name’, ‘hospital’, ‘bed number’ to calculate the actual demands of patients and detect the contradictions between plans and actual situation.
Fig 1: Time trends of total health expenditures and out-of-pocket expenditures in China[1]

2.Better awareness on mental health

As mentioned in SDG 3.8, promoting mental health and well-being need everyone’s attention. As shown in Fig.2, the number of suicides in US is increasing and going to exceed 45000. The stress comes from all sides, we can’t predict and sometimes hard to prevent. In addition to real-life consult, what kind of role can SMA play in keeping mental health?

Fig.2: Total suicides in the US, 1981-2016[2]
  1. NLP and sentiment analysis could be vital approaches. The content of Twitter, Weibo posted in public could be classified as positive or negative. If an account keeps posting negative content for about 10 days, the social media platform could firstly try to pop up some relief post like cat movies or relaxing songs.
  2. With the help of SMA, sensitive words like ‘suicide’, ‘die’, ‘testament’ or other negative words could be quickly detected. Experts could try to comfort through private massage or comment, and call on the police if necessary.

3. SMA can also….

a) Infectious disease control: Through statistics of shared locations, find people most likely to be infected. For example, if someone shared location in ‘华南海鲜市场’ in Wuhan among December to January, he/she could probably go there and has a great chance to be infected of novel coronavirus. SMA could help to track those people and control at the first time.

b) Help quitting smoking and alcohol: pop up public service announcements or relative short videos to the users who posted about smoking or alcohol with high frequency.

And many other aspects you can share with me.

Above suggestions are my primary assumptions, the balance between privacy and SMA still should be noted. This blog is a little bit long, but every aspect is coming from deep thought. Hope it can give you different insights.

[1] Time trends of total health expenditures and out-of-pocket expenditures in China. Nabholz, Christoph. (2017). Diabetes in China.
[2] Suicide in the United States. Wikipedia. Retrieved from

Good health and well-being — Sustainable Development Goal 3

Why health matters?

Good health is crucial and essential to sustained economy, social development and poverty reduction[1]. Good health is nearly the basis of development. First, tight relationship exists between health and economy status of a country. When the birth rate increased by 10%, the annual economic growth rate was about 0.3%-0.4%[2]. Besides, heath does a great impact on education. In[3], the average math and reading achievement of children are negatively affected by several health conditions.

Healthcare. Credit:https://www.getfoundquick.com/5-reasons-why-seo-is-important-for-the-healthcare-industry/

Brief introduction of SDG3—Good health and well-being

Even though the medical level has a great improvement all over 21th century, there still exists difficult medical problems and several places are still in poor medical conditions:

  • Only half of women in developing regions receive the recommended amount of health care they need[4]
  • 940 000 people died from AIDS-related illnesses in 2017[4]
  • Four out of every five deaths of children under age five occur in Sub-Saharan Africa and Southern Asia[4]
  • An annual global age-standardized suicide rate is 10.5 per 100 000 population.[5]

SDG3 is mainly aiming to the following parts:

  • Reduce maternal mortality and neonatal mortality rates
  • Reduce communicable diseases like AIDS
  • Reduce the injuries/deaths from road traffic or hazardous chemicals
  • Strengthen the prevention and treatment of substance abuse
  • Strengthen sexual and reproductive healthcare services
  • Strengthen development of vaccines and medicines
SDG3 targets. Sources:http://www.euro.who.int/

Stay alert !

In addition to government policies and UGO efforts, what can we do in our normal life to help achieve SDG3 and maintain our body health?

  • Get regular medical check-ups
  • Get vaccinations and use condoms
  • Wash hands and wear masks in flu season
  • Quit smoking and drinking
  • Periodic disinfection of clothing and other items
  • Talk to someone when you are depressed, don’t be afraid of psychiatrists

Maintaining is better and easier than curing, we should stay alert to take care of ourselves. However, honestly, we can never ever avoid the happening of diseases in our life. Serious or small diseases would always occur, more importantly, we have to learn how to live with diseases and keep positive.

Stay united

When Ebola virus outbreaks, WHO organized doctors and nurses from all over the world to help the people in Congo; when novel coronavirus occurs in China, a huge number of medical supplies are donated from other countries and the vaccination is being developed with the help of the researchers around the world. The unsolved health problems are not just for a person or one country, it’s for the whole planet. The world is a huge circle, every part of it matters. Faced with diseases, we are incredibly small and fragile, but also, we are unexpectedly strong.


What is social media analytics and why do we need it?

In the age of information explosion, social networking is twisted from the traditional media to the modern social media, followed by an emerging topic——social media analytics.

Fig1. The change of social networking [1]

What is social media analytics?

According [2], normally driven by a specific target, social media analytics is basically informatics tools to extract, analyze and summarize then visualize the information from the collected data of social media. We can summarize the specific steps as follows:

  1. Put up with a problem
  2. Identifying the available data of social media
  3. Data collection: use web crawler or API tools.
  4. Analyze the data: some machine learning algorithms would be implied
  5. Present the results: summarization and visualization.
  6. Draw the conclusion

We can use an example to simply explain this. Someday a video become popular in a sudden——Chinese singer Cai Xukun is playing basketball in a strange way. After being reposted by 10 thousand times, the number of Cai Xukun’s followers is exceeding 20 million on Weibo. And we want to find out if someone is manipulating behind the sudden increment.

Fig.2: An example of specific steps to do social media analytics

Why social media analytics is important?

From business perspective

For companies, sentiment analysis in social media would help them improve their products and adjust marketing strategies. Nowadays, people usually share their product or food experience in Instagram, Openrice or Xiaohongshu, etc. The companies could do sentiment analysis on their posts and see what people think of the products to improve their services or change marketing ways.

Fig.3: Sentiment analysis engagement for a leading food service firm [3]

For customers, we can use social media analytics to detect the potential advertisements. As social media takes over our lives, some commercial advertisements could be hidden under the daily posts of some Internet celebrities. According to [4], there’re 60% of users know new products through Instagram. In Fig.4, up to 5k sponsored posts are sent in an hour on Instagram. To determine the true recommendations, you may need social media analytics.

Fig.4: Sponsored posts on Instagram by time of day [5]

From political perspective

Social media analytics could help to avoid the potential political influence. During the three debates of 2016 U.S. Election, Twitter manipulation could be involved through some highly automated activities. Highly automated accounts generated 4 pro-Trumps tweet to every pro-Clinton tweet in the first debate.[6] Besides, through sentiment analytics, we could predict the election results by gauging people’s opinions on social media and plan for the next step at the first time.

Fig.5: Total hourly Twitter traffic around voting day, 2016 [6]

Thank you for reading this blog, hope it can give you some different insights.