About the controversial Naive Bayes Classifier

Concerning sentiment analysis, machine learning techniques makes it more convenient. Naive Bayes is undoubtedly one of the simplest classifiers in supervised machine learning due to the simple mathematics involved and its easy coding feature with programming language.

In our lecture, we already knew that Naive Bayes classifier  is a simple probabilistic classifier which is based on Bayes theorem with strong and naive independence assumptions. It is a common word classification technique which is widely applied in filtering spam email or sexual contents, categorizing specific documents or pieces of text, even identifying sentiment of the writer.

Photo from : http://machinelearningmastery.com/naive-bayes-classifier-scratch-python/ (Photo by Matt Buck, some rights reserved )

But Naive Bayes is often criticised for its oversimplified assumption for its calculation that  probability of each attribute belonging to a given class value is independent of all other attributes. You can say its independence assumption is its weakness, but without this assumption, it is not possible for us to obtain a fast and effective result. It is also worthy noting that it is often used as a baseline in many researches due to its outstanding performance in some cases.

Again, no single solution can solve all problems. Whether the result is good or not, it depends on whether we use a suitable algorithm to solve the particular problem and how we use the technique involved to solve more problems as every method has its strength and limitations. Especially when we only have limited memory or CPU, efficiency of Bayes Classifier makes itself an excellent choice. In Naive Bayes Classifier only need a short period of time to do the machine training to obtain the result. As a result, it is very handy. Do you have any opinions towards this controversial algorithm? Do you think Naive Bayes Classifier outperform other algorithms?

Reference:

1. Prof. Rosanna Y.-Y. Chan(May 2017). Week Seven (1): Social Media Analytics: Sentiment Analysis [Powerpoint Slides]. Retrieved from : https://elearn.cuhk.edu.hk/bbcswebdav/pid-1981443-dt-content-rid-5769767_1/courses/2016R2-IERG3320/IERG3320_2016-17%20Term%202_Week7-1-updated_1March.pdf
2. Vasilis Vryniotis(Oct 2013). Machine Learning Tutorial: The Naive Bayes Text Classifier. Retrieved from: http://blog.datumbox.com/machine-learning-tutorial-the-naive-bayes-text-classifier/
3. Sebastian Raschka(Oct 2014). Naive Bayes and Text Classification.Retrieved from:
   http://sebastianraschka.com/Articles/2014_naive_bayes_1.html
4. Sunil Ray(Sep 2015). 6 Easy Steps to Learn Naive Bayes Algorithm (with code in Python). Retrieved from: https://www.analyticsvidhya.com/blog/2015/09/naive-bayes-explained/

Language and Body Language

In terms of language, you may immediately think of words in different kinds of language. But why we understand the language and how does the language have its meaning? Just like a Chinese word, if it has different combination with other words, it can have a totally  different meaning; if the sentence uses different symbols to compose, a different manner of speaking will be resulted. Thus, to understand the  language functions, we should familiarize ourselves with the SIGNS and SYMBOLS. Signs include icons, indexes and Symbols.

  In our daily life, using emoji is also a common way to express our feelings as a kind of language which is called visual information. This kind of information can be directly interpreted by us even though no one has explained it to us. Thus, I think the form of language is enriched with the development of the Internet.

Moreover, another topic in LANGUAGE is body language as body language often helps use to stress our words as an assistive tool whenever we communicate with others. However, unlike WORDS, body language may not have a specific way to express. Different movements can be interpreted in a very different way by other people. And there is an interesting article mentioned that men and women may have different kind of  body language to explain the same feeling, so we may need a different body language decoder system for men and women. So what do you think? 

Article Source:

http://www.cosmopolitan.com/sex-love/advice/g3961/body-language-decoder/

More About Social Psychology and Human Cognitive Processing

Social Psychology in Human Information Interaction and Social Media is a very interesting topic and it talks about the psychology behind those information actions like communication and information sharing. What are your intentions that make you want to share information with others? How do you use online social media everyday? Just take myself as an example, I am used to share interesting daily encounters with my friends because sometimes I want to express my feelings about it(self-expression), receive feedback from friends or just want to share the joy with others. As we all know that ‘grief shared is halved, joy shared is doubled’.  This kind of social actions can be exactly used to explain Reciprocal Determinism concept we have learnt. If me as the Person, who express my own belief and share joy with others (Behavior) externally to the social media environment as online opinions can influence the belief and behavior of one another like making one another happier. And I in return feel satisfied if others become happier after reading my sharing.

source from: http://teachingadolescents.weebly.com/bandura.html

Apart from the social psychology behind social interactions, I hope to clarify what the three levels of Human Cognitive Processing mentioned in both week-1 and week-3 notes are as I am still confused by difference between Cognition and Meta-cognition. The three levels include: 1st level- Cognition, 2nd level- Meta-cognition and 3rd level-Epistemic Cognition.

Cognition means individuals compute, memorize, read, perceive, solve problems(those actions are so called first-order tasks); Meta-cognition means individuals monitor their own progress when they are engaged in these first-order tasks; And Epistemic cognition means individuals reflect on the limits of knowing, the certainty of knowing and criteria of knowing. To explain it in a simper way, let’s take solving a mathematics problem as an example. Solving the problem comes from our cognition, meta-cognition will then regulate our cognitive process, for example, realising the strategy we are using is not working and trying another aaproach. And finally epistemic cognition will help to reflect on this and transfer all these strategies and errors into our knowledge and knowing. I hope all these will be helpful to you too 🙂

Reference: http://www.cie.org.uk/images/272307-metacognition.pdf

 

What is Human Information Interaction and why is it important to me?

Due to the mobile technology revolution, there are dramatic changes in our daily communication as well as social networking. Information is no longer solely provided by traditional mass media like newspaper, television, radio or books. Thousands of Internet and mobile users can also be the content creator or collaborator through online social media. In short, if you are an Internet user, we all contribute to the Big Data. The relationship between human and  those incredibly large volume of User Generated Content(UGC) information is exactly the utmost importance of HII. The way we interpret the data and the way we explain human information behaviour requires a multidisciplinary system to explain, and areas may range from human-computer interaction(HCI) to computer-supported cooperative work, human factors, social informatics, management, and library and information science (LIS). But why HII is so important to us? Apart from understanding the characteristics of human beings, HII also provides a way to test theories of cognitive architectures in research field while models could be built to predict effective information design.[1]

We may have heard that Big Data is a very power tool nowadays in digital marketing to enhance user experience and etc. But someone may think differently that “Big Data is too big[2]” and is not a good strategy to use Big Data as the lack of the right data at the right time leads to being faced with difficult decisions at times when they might not have access to the information that they’d like. So what do you think?

 

Reference

[1] HLWIKI Canada contributors, “Human information interaction (HII),” HLWIKI Canada, , http://hlwiki.slais.ubc.ca/index.php?title=Human_information_interaction_(HII)&oldid=

143971 (accessed January 27, 2017).

[2] MICHAEL LAPS, “Big Data has no role in digital marketing”, Mumbrella, https://mumbrella.com.au/big-data-has-no-role-in-digital-marketing-422319  (accessed January 27, 2017).