In class, we talked about algorithm for web engines. We also discuss about how there are many informations in the web and how to figure out what you really mean when you typed “Cornell”. We skimmed about the information about clustering of the information and how Amazon delivered the information to you by recommending information by suggesting in “Customers Who Bought This Item Also Bought”, “What Do Customers Ultimately Buy After Viewing This Item” and “Looking for ‘Topics’ Products?” sections.

In the IEEE Spectrum Issue March 2008, there was an article that talked about the clustering the newspaper article by suggesting newspaper article to reader around the world. The main purpose of this article is to describe different algorithms used by the different providers of the Internet news. However, the article also brings insight to different issues that was discussed in class. For example, this article talked about the world is changing with every hours, every minutes. However, newspaper just provide a series of articles for one day without any change in the front page. Therefore, in order to reflect this changing world, the article stated how it might be a good idea to turn to internet to find newspaper articles.

The task of clustering news is a hard problem to solve. This is harder problem to solve than search engines because now they’re not keywords but events. How do you define events? Amazon and Netflix tried to give recommendation based on your interests and your previous searches. Therefore, using this idea bought about some idea on the algorithms for clustering news. Google News have some algorithms that tried to calculate the likelihood of the article of reading other articles given one specific article and also tried to calculate what each reader’s top stories should be. There are also more complicated algorithms where Google News tried to analyze the contents of the articles based on phrases or keywords and then complied into a tree.

Besides Google News, the news recommendation program such as Findory exist. Findory recommends news based on the news you click on. For example, if you click on the article on Microsoft, then if might recommend news articles on technologies at Microsoft or it might brings up information on Microsoft buying Yahoo or it might brings up some other tech companies such as Apple. This recommendation program in lower-hierarchy is like a social network of people who like the same article. Although no one knows about this network existence, this is how the news are clustered. This is just another way to bring about information sources to people.