Use Information Technologies to Reduce Inequality

My group topic is about reducing inequality. Apparently there are many aspects of inequality. What I have introduced in the last blog post are education inequality and gender inequality, which are still very serious issues within and among countries. So how can we use information engineering technologies to solve these problems?

1. Information technologies reduce the education crisis in Africa

Many students in Africa cannot receive education well because of the poverty. As the most undeveloped area in the world, the backward economy leads to the lack of educational fund, which further causes bad teaching environment, and the lack of teaching equipment, books and reference materials.

Tough conditions for education in Africa. Source from https://www.ladiestrekking.com/africas-education-crisis-in-school-but-not-learning/

Distance education is a good measure to solve this problem. Benefited from computer network technology, multimedia technology and big data technology, distance education breaks through time and space constraints, which means that students in Africa no longer have to worry about the education resource scarcity.[1]

Students participate in an online course. Source from https://cs.stanford.edu/people/eroberts/cs201/projects/2004-05/distance-education/201finalproject_files/page0005.html

2. AI finds gender inequality at work

A company named Kanjoya is applying NLP to address gender inequality at work. Its NLP algorithm can figure out what people are really thinking when they fill out the employee survey by analyzing the tone and context and then learning the features from the information. [2]

The company which use NLP to combat gender inequality at work.

Kanjoya can aggregate all the sentiment information received from the employee survey and then analyze the data. The result, like demographics, allows people to find out the potential problems in the company, including gender. If there are so much negative voice from women, the sentiment data proves that women are not as fair as men in this company. Another sign which shows inequality between gender at work is that more leadership evaluations to men and less to women when they doing the same kind of job.

Here is a video which shows how we can measure and combat gender discrimination in the workplace for your details. It can help you have a clearer view on how Kanjoya apply NLP to address this problem. [3]

Measuring and Combating Gender Biases in the Workplace from Seth Grimes on Vimeo.

 

Reference:

[1] https://cs.stanford.edu/people/eroberts/cs201/projects/2004-05/distance-education/201finalproject_files/page0005.html

[2] https://www.fastcompany.com/3052053/how-artificial-intelligence-is-finding-gender-bias-at-work

[3] https://vimeo.com/138071811

Reduce Inequality

Although we always emphasize the equality of rights, rules, opportunities and income distribution, the goal of total equality has not been fully realized. We are now living in a world where inequalities are everywhere. In order to reduce inequality within and among countries, the United Nations has proposed the development goal to transform our world to a more equal society, which attracts wide attention around the world.

Reduced Inequalities, one of the UN sustainable development goals. Source from http://www.un.org/sustainabledevelopment/inequality/

I will talk about two problems and the way we can reduce the inequality below.

Education Crisis in Africa

Among 128 million school-aged children in sub-Saharan Africa, only half of the them can receive education, learn skills and live healthy lives. The great inequality among sex, income and local education can be seen everywhere in large parts of countries which lie in sub-Saharan Africa. In this matter, the cruel reality cannot get away from poverty in African regions. Teachers are more willing to work in cities for more opportunities, higher salary, better quality of life instead of that of the poor region so that they can make better use of good infrastructure, medical care and some other services and public goods. Besides, the economic condition in poor regions causes negative impact on the quality of the education. Many students even leave school and come out to work at a very young age just because of their illnesses and their parents’ illnesses. They need money to cure diseases. As we can see, the phenomenon of unfair education in African regions seriously hinders the development of their society and economy.

Poor educational facilities in Africa. Source from https://www.buildon.org/about/the-education-crisis/ and http://www.worldvision.org.hk/en/learn/education-crisis

In my opinion, to face the education crisis, the local government should publish corresponding measures to address this problem. They need to optimize the infrastructure, sanitary conditions and medical conditions and promote the modernization of agriculture. Besides, stepping up urbanization construction is a great strategy that promotes the development of industrialization and economic growth, and narrows the difference between urban and rural areas in the quality of education. The African government should take actions immediately to collaborating with other countries to improve teaching standards and quality, otherwise the talents of thousands of younger generations in Africa will be ignored, and the economy and society in Africa will be more and more depressed.

Reduce Gender Inequality in Life and at Work

Gender inequality mainly refers to the inequality of the women social status in the society. In ancient times, men dominate the society because of the advantage of physical strength in agricultural civilization. However, when we are in the age of industrial civilization and information, the development of society mostly depends on our brains. Since there are almost no differences in intelligence between men and women, women’s position will be definitely improved and gender equality will finally be achieved, which is the inexorable law in social development.

Workplace Gender Equality. Source from https://www.forbesmiddleeast.com/en/u-a-e-leads-for-workplace-gender-equality-in-mena/

However, although some measures have been taken to minimize the gender inequality, they are not sufficient. In the traditional concept, women are viewed as weaker than men. In many rural areas in China, the misconceptions that men are superior to women is still dominating many people’s minds. Thanks to the One Child Policy, families prefer boys than girls, and due to this out-of-date concept, there has been female infanticide on a vast scale.

Source from https://www.youtube.com/watch?v=eNKQT7Ub2Ps

In work places, although 40% of the global work force are women, they are paid much below than that of male workers when doing the same kind of job, despite being equally knowledgeable and skilled. One study has shown that Japan will lose 15% of its GDP if the problem of gender discrimination is not addressed in the future. Beyond the economic costs, gender discrimination can also cause huge loss to not only the individual but also the society. Although more than half of the population in the world are females, only 7% of them are government leaders. If we want to further promote gender equality, women equal voice should be emphasized in public life, otherwise the decisions and policies which made by the governments are more advantageous for men.

From my point of view, the achievement of gender equality needs combined efforts of men and women. In China, I think men should take the same(or more) responsibility in family as women. Why? Because on average women spend more time and efforts on the family and bring up their children so that they may sacrificed their own career. If men and women share the same responsibilities in family, women will have more time and energy to develop their own business. Fundamentally, we must respect for women from the bottom of the heart. Only by doing these, we can reduce the gender inequality and improve the position of women. Recently, the One Child Policy has been changed to the Two Child Policy. I think this will be a good start in the process of pursuing gender equality in the future in China.

Reference:

[1] https://www.brookings.edu/blog/up-front/2012/09/17/africas-education-crisis-in-school-but-not-learning/

[2] https://borgenproject.org/10-facts-africas-education-crisis/

[3] http://bold.expert/a-new-approach-to-africas-education-crisis/

[4] https://en.wikipedia.org/wiki/One-child_policy

[5] https://www.summer.harvard.edu/inside-summer/gender-inequality-women-workplace

[6] https://www.huffingtonpost.com/nake-m-kamrany/gender-inequality_b_1417535.html

My Viewpoints on Semantic Analysis

Semantic information, refers to what we have learned in previous lectures, is the meaningful information which is provided by languages, words, images, data, symbols, etc. It has domain characteristics, and there is no semantic information that does not belong to any domain. Thus, when we consider the semantic information, we need to put ourselves into the specific context. It means that the same object may have different meanings in different environment.

For example, the sentence “It is good to have a sunny day.” has two different meanings in different environment. If we are exactly in a sunny day, this sentence means that we admire the good weather and we really enjoy the weather that day. However, if we are in a bad weather, this sentence expresses that we hate the bad weather and miss the good weather so much. Let’s take another example. If we just say “orange”, the people who is not next to you may find hard to identify whether you are talking about the color or the fruit.

Figure 1 – Ogden and Richards (1923) meaning triangle (source from http://iq3group.blogspot.hk)

It shows how significant that we need to combine with the specific language context when we are analyzing semantic information. For example, the picture below is a classical Chinese poetry which was wrote by Li Yu. By combining historical background, Human can understand what this word or this sentence actually means well according to the context. And we can also see that the Chinese semantic platform cannot handle the poetry well.

Figure 2 – (source from http://www.gushiwen.org/mingju_1496.aspx and http://nlp.qq.com)

But with the rapid development of AI and NLP, content analysis and semantic understanding are not just human’s ability. Machines can also do well in these fields. Recently, a team named iDST led by Alibaba beat the highest human score with an 82.440-82.304 victory and broke the world record in reading comprehension in a top competition named SQuAD, which was held by Stanford University. The SQuAD competition constructed a large scale of dataset of reading comprehension, which includes more than 500 articles from Wikipedia and 100,000 questions aiming to test whether these machines can give correct answers or not after processing a large amount of text information. As we can see that, nowadays the machines can even do better than our human in some kind of fields when dealing with the content with complex context.

Figure 3 – (source from https://rajpurkar.github.io/SQuAD-explorer/)

When doing sentiment analysis, we need to figure out whether the word is positive or negative. This is very important. It is not so hard for human, but it is hard for machines to do that kind of thing. Suppose that the analysis object are the comments of the iphone 7 Plus, as we can see in the picture below.

Figure 4 – (source from https://www.amazon.cn)

English has already had great dictionary resources, like SentiWordNet, which does well in analyzing the positive, negative, subjective and objective words and the emotional intensity of the words (If you are interested in emotional analysis, you can visit http://sentiwordnet.isti.cnr.it/ for more information). In the Chinese field, we also have some dictionary resources for sentiment analysis, like Hownet and NTUSD. When we want to identify whether this sentence is positive or negative, we need to match the words in the sentence to the dictionary and compute the sum of emotion. However, different fields have different emotional words. As you can see in the picture above, the words “blue screen” and “brick” are not the common emotional words in the dictionary, but they obviously express the feeling of dissatisfaction. Therefore, when we are doing sentiment analysis, we are actually doing opinion mining. We need to find out all the attributes of the product, take an example of the phone, screen, battery and camera are all its attributes. We need to firstly mine the attributes of the product and then analyze the relative emotion of the attribute. After analyzing the emotion of all the attributes, we can summarize them up to form a customer’s evaluation of a product.

There are quite a lot research working on content analysis and emotional analysis. I think the development of semantics analysis will make our life more convenient and wonderful in near future.

References:

[1] https://blackboard.cuhk.edu.hk/bbcswebdav/pid-2376210-dt-content-rid-12064746_1/courses/2017R2-IEMS5720/IEMS5720%20Social%20Networking%202017-18%20Term%202%20-%20Lecture%205.pdf

[2] http://iq3group.blogspot.hk/2012/11/tower-of-babel-semantics-initiative-and.html

[3] https://rajpurkar.github.io/SQuAD-explorer

[4] https://www.cnblogs.com/arkenstone/p/6064196.html

[5] https://en.wikipedia.org/wiki/Semantics

 

What is information and what does it mean to me as an information engineering student?

Information is no longer unfamiliar to us all because we are already in the information era. In the primitive times, human beings got information from their gesture and some hieroglyphical characters printed with the stone. Soldiers built beacon towers to transform the battlefield information or give the alarm using fire and smoke. Lovers with long-distance relationship write letters to each other to show their love in old days. And nowadays we can get information from anywhere such as books, TVs and the Internet. Although information has different formats in different time, information is ultimately the stuff that we can distinguish things and learn knowledges from it so that we can know the world and reform the world.

As an information engineering student, we are naturally always dealing with a large amount of  information. In social networking course, we learn that information is the source consists of a finite set of information bits(see picture 1).

Picture 1. Source from blog.sina.com.cn

It can be coded and decoded in the specific format, which depends on its purpose. For example, if you want to call someone, the voice is the information, which is processed and transmitted by the transmitter, goes through the channel and finally received and decoded by the receiver. In 1948, Shannon proposed the concept of entropy, which can be defined as the formula below:

Picture 2. Source from http://kevinmeurer.com/a-simple-guide-to-entropy-based-discretization/

The larger the entropy, the more information content a message contains.

With the development of the Internet, information becomes more and more important not only for our users but also for the companies. Especially in today’s big data era, data(also the information) is the king. Almost every Internet company nowadays tries to collect users’ data to track their behaviors or train their algorithms for machine learning. Social networking provides the way for us to increasingly observe the complex behavioral patterns of the human beings from the information. The core value of the information is that people can mine the useful data, analyze the behaviors, habits and preferences of the users to create products and services more suitable for the users.

With a large amount of facial data, the machine can learn facial features and detect the face in photos. The new iPhone X does pretty well on face recognition for unlocking the device and mobile payment. It actually uses the information of our faces, like the geometric characteristics of our eyes, nose, mouth and the distance between them(see picture 3 and 4), to learn the features so that when the camera on the device see your face it can recognize you by analyzing whether your face’s features is matching the features in its database.

Picture 3. Face recognition on the iPhone X. Source from https://www.macworld.com and http://adage.com

Picture 4. Information in the face. Source from http://www.aibang.com

Also, apps like Taobao and Jingdong can automatically recommend goods which are fit for users’ interest after collecting the trace of browsing and analyzing these data in a very short time. So our behaviors on the Internet, like shopping, searching, and even the content of what we chat, have diverse information to show who you are, what you favor, etc(see picture 5).

Picture 5. The recommendation system of Jing Dong. Source from chinaz.com

All in all, I think information is the wealth in today’s society. The company who obtain more information will be more competitive in the business field and have greater opportunity and competition to success in the future. So as an information engineering student, we can not be flooded with information and need to fully exploit our technology abilities to discover the useful information around us for data mining. Meanwhile we need to learn more knowledge to be more competitive in today’s information era.