Due to the vivid explanation of our teacher， I had got a preliminary impression on social media. Just as the name implies,social media means that everyone of us can participate in spreading and sharing information,which breaks the old pattern of messages spreading.
We had learned a lot of notions of social media and how to analysis contents.From my perspective,the K-means clustering is the most appealing part.During the class ,we learned the basic notion of K-means clustering,and using vector space model and cosine similarity to calculate estimate the distance between documents.
The procedures of the algorithm are as follows:
1.Chosing K data points from all data points randomly,each of which stands for a centroid.
2.Allocating the other objects based on the distance to the central points.
3.Calculating the average of each cluster and getting the new centroids.
4.Repeating step 1-3 until the function is astringed.
Despect the fact that K-means clustering algorithm is of great use in data mining,it is effective only when it is used to handle numerical value data.
When it comes to data like colors,shapes,stripes ,it will be failed.
At last,I would like to say that "Social Media Analysis and Development" is a really interesting course,and I will definitely learn a lot of from it.
Thank you very much for your sharing ，Especially about the procedures of the algorithm is very easy to understand.From the blogger, it can be seen that You have a unique insights on this course"Social Media Analysis and Development".
回复删除found that you read some reference in your blog, and you also found some interesting perspectives which were not mentioned in the lectures. I like your article very much.
回复删除so clear definition, Which is really useful. I agree your point . By the way ,what is your opinion of economy effect of social media? Does it necessary for the existence and development of social media?
回复删除Reading your blog helps me learn more about K-means clustering. I also got very interested in K-means. However, it should be no wonder that we have not mastered the understanding of them all. So can we discuss this offline?
回复删除You not only express your idea about social media, but also you have a unique cognition in term of analysis contents. The graph which your draw by k-means is attractive. I do admire the way of your thinking.
回复删除Got a deeper understanding of k-means cluster. Even algorithms can be very impressive and interesting in this field, so maybe I will further research in NLP and SNA.
回复删除Hi Wenhan,
回复删除You've made a good summary of the K-means algorithm which is a very basic and widely used clustering method. I also learned about the limitations of this algorithm. However, I think when dealing with non-numerical values,maybe we can find some way to quantify it before perform the clustering method. As far as I know, this algorithm can also be used in image processing. Anyway, thanks for your sharing.^.^