Recently I read a blog entitled “Predicting your favourite TV Progarmme”. This article introduces the program recommendation system of TVB.
There are two steps for the system to prepare a list of personalized video recommendations. The first step is the candidate generation stage, where thousands of videos can be reduced to hundreds. The second is the ranking stage, where the surviving videos are ranked in the order based on a classification neural network. Finally, the top N will be recommended to users.
About the recommendation system, there are four main steps before the recommendation results are generated.
- Step 1: Embed programmes into a vector representation.
- Step 2: Prepare the dataset.
- Step 3: Train model.
- Step 4: Tweak the model.
I will not go into the details of the model building process here. If you are interested, you can go to this website(http://tech.tvb.com) for more information. But for those who do not have a background in machine learning, that article may be difficult. I suggest that the readers first learn some programming knowledge about machine learning, and then understand the mechanism of TVB recommendation system in detail.
Jonathan's Blog 
Ahhhhhh, thanks for your suggestion to learn machine learning first which is quite realistic. After looking through your whole article we can find that you have a clear understanding of personalized video recommendations and recommendation system, and you list four steps before the generation of recommendation results which greatly helps the beginner to have a rough cognition.
赞赞
Short and fine. Thank your again for your suggestion to learn machine learning first. It is really the basic knowledge that we need to master before giving comments on recommender system. Your blog has covered the whole process of how a typical recommender system works. But I think any conclusion should be supported by material and your blog is kindly in lack of that. What do you think about that?
赞赞
Actually the blog is hard to understand for novice to understand that blog. However, it tried its best to make the content clear and easy to understand. Containing some many techniques in machine learning and social media analysis, that blog manage to explain the ideas concretely and vividly. To this point, I think it’s still a friendly one to beginners.
赞赞
Well, the blog is short and generalized. You introduced how a recommender system works to us. Actually, for those who are not familiar with machine learning, you can talk about other things to help them understand it. Anyway, thanks for your sharing.
赞赞
Hhhh, I totally agree with your opinion that we should have a background about machine learning before we learn something about recommend system. I will follow your advice. Hhhhh. Actually I read the same article, maybe we can exchange ideas about that.
赞赞
thanks for your conclusion on that blog. I haven’t read that blog yet and you have clear state four step on how TVB doing recommendation system. Actually I would like to know why they got this four step instead of the traditional one. the problem they were facing and the way to solve. That will be great if these information was included in your blog.
赞赞
After reading your article, I have a clearer understanding of TVB’s recommend system. The four steps before the recommendation results are generated vividly introduce how a recommender system works.Thank you for your nice job.
赞赞
Thanks for sharing! Your article is short and clear. You give a conclusion about how the recommender system works in the TVB. And you suggest that before we understand the recommender system, we should grasp some basic machine learning concepts. I agree with you!
赞赞