With the development of data science and data mining, some consumer-oriented constitution such as TVB also want to make to use of big data analysis to provide better service to consumers. As over thousands of hours TV content are ready in content library, some of them may not be easily discovered by user. Recommendation technology is an ideal way to provide personalized content to user and encourage user to enjoy our service.

In early times, TVB used collaborative filtering approach to make the predicted recommendation by creating a user-to-item matrix and using SVD (Singular value decomposition) or NMF (Non-negative matrix factorization) to find the similar users or similar items. However, this method can’t predict effectively because they don’t consider the sequential information between services. For example, if a user watched TV programme “The File of Justice I” , then a conventional collaborative filtering may rate “The File of Justice II” and “The File of Justice III” equally as likely. However, we all know that the user will watch “The File of Justice II” before “The File of Justice III”. This requires our model to learn the sequence of the users’ watch behavior.

As the result, we improve the methods using machine learning algorithms to build a two-step approach. Firstly, in the candidate generation stage, thousands of videos available are narrowed down to couple hundreds. Next, in the ranking stage, the surviving videos are further ranked in the order based on a classification neural network. Finally, the top N of the ranked order will be used to serve users.

Step 1: Embed programmes into a vector representation

Word(vector) embedding is an effective way to represent the video itself. Before we feed the sequence of video into the model, we must first transform it into some numerical representation, i.e. Embedding. For example, “Game of Thrones” will be transformed into [0.1, 0.3, -0.6]. A benefit of using the word embedding technique is that it can group videos into similar neighbours without us having to specify or label anything.

Step 2: Prepare the dataset

We casted the recommendation problem into a classification problem. The idea is to train a model to predict if one program is good to be recommended given the view history of a user. Each of the P1, P2… below are the vector we created at the last step. You may just flatten the list of vectors or pass it through an RNN/LSTM layer before further process it with a classification model of your choice. Here, let me just skip the details of that and focus on the training data set preparation.

Step 3: Train model

You can use your choice of the model for classification. There are some of the common algorithms.

Step 4: (Optional) Tweak the model

While we all feel uncomfortable when we add additives into our model, it is good to have tuning mechanism, to make the recommendation results look more like the ones in the business owners’ mind.

Let’s get back to the preparation of training sample. Remember we created this set of training sample from user view histories.

Now, let’s assume P3b, P4b, P5b are the behind-the-scenes of P3, P4, P5 respectively, and the management wants it to be recommended too. In order to tweak the model, we can introduce some additives to the model in this way:

These samples will tell the model that it is good to recommend P3b if it sees a P3 in the last position of the view history.

Conclusion and view

The boosting development of big data analysis technology has helped and driven the traditional consumer-oriented companies work better. They can use more flexible approach like recommendation systems and NLP to provide better services, which can help more companies get to know such technologies and drive the whole industry ecology to be better.

Reference

Predicting your favourite TV Programme

https://medium.com/techattvb/predicting-your-favourite-tv-programme-eda31a5e51b3

在 Medium.com 上查看

How I exercise SEC at my cognition when seeing classmates’ comments?
when we are engaged in knowledge-related activities with the presence of the others, we are always influenced by others’ cognition. Originally, we hold our opinions and write them down to share to others, that means we also want to deliver our cognition and thinking to others. So when we receive others’ comments about our blogs, we will take them into account to our cognition.

Firstly, we will judge whether these comments are positive or negative to our cognitive. If they are positive, we may take into account and furthermore compare them with our original thinking. If they are negative, we may quickly discard them.

Secondly, when confirming the positive comments we will compare it with our inherent thinking and try to find some connections between them. Specifically, we are trying to add them to our knowledge graph. These positive comments will be part of our knowledge graph.

Finally, these new part of our knowledge graph my drive us think more about the project. Many of the parameters of the graph may be updated, and we will get our cognition updated. Besides, we may want to share our new thinking to others, and update our different thinking together to gain more knowledge.

Social Presence Theory is defined by the different physical proximities produced by different mediums, the two more popular being face-to-face communication and online interaction. Social Presence is measured by the ability to project physical and emotional presence, and experience it from others in interactions. media differ in the degree of social presence, they allow to emerge between two communication partners; and the higher the social presence, the larger the social influence that the communication partners have on each other’s behavior.

Social Presence Theory plays as an important role in social analysis. Due to the different degree of social presence, we can classify different social media content by these degrees. In the recommendation systems, especially in social network, the system designers can divided people into friends or stranger or someone else by there presence; and define the familiarity by the content type(face-to-face or telephone conversation), and the synchronous or not(live chat or e-mail).

More and more data analyzer are going to use different social psychology theories in data mining and business analysis. In these regions, social presence theory can help them to do classification and stratification, thus effectively comprehend the massive social media data and explore more latent relationships between people.

Popper’s Three Worlds Ontology

One of the most popular epistemology models (except for perhaps in the behavioral sciences) is Karl Popper’s writings on the Three Worlds of Knowledge. The behavioral sciences (knowledge/learning/management professions) seem to prefer and stay within the realm of Michael Polanyi’s concept of personal and tacit knowledge. However, Polanyi’s epistemology is narrower and has a limited basis for understanding knowledge as compared to Popper’s work, which provides a broader epistemological foundation.

In his Objective Knowledge (1972), Karl Popper introduced the idea of three ontological worlds or domains. The first world is the world of material objects, events, and processes, including the domain of biology. The second world is the world of mental events, processes, and predispositions– the world of beliefs and other psychological phenomena. The third world is the world of the products of the human mind.

What dose social media belongs to?

In my work in Social Media Analysis, I’ve relied heavily on Popper’s three worlds ontology and also on his ideas about knowledge. I think Social Media is belongs to the interaction between World1 and World2.

First of all, only when based on the material objects and events in World1, such as people’s events or world incidents, can the origin of social media happens. These are all the basic content of human mental. Secondly, social media contents can be produced, interacted, transmitted all over the network, and then to people’s mental events. On these conditions can mental events be further processed and predisposition among the mental world.