VirgoLab

Many types of recommender systems exist such as non-personalized, demographic, content based, content based, collaborative (user based), collaborative (item based) and model based. Item based collaborative models have been applied successfully in commercial settings thanks to scalability and quality advantages as compared to others. Model based approaches differ from the rest which rely on memory of events. Instead they involve the creation of a probabilistic, decision tree or neural net model that attempts to identify the underlying logic to users’ choices.

Sparsity, Scalability, Cold-start, Implicit Ratings, Dealing with Multiple Criteria, Context-dependant Recommendations.

我仍然尊重記者必需的專業與訓練，許多「公民記者」的撰述並不能成為獨立而平衡的報導（當然，很多專業記者也做不到），但卻提供了另外一種觀點、或是突破會讓主流媒體大招牌卡住的限制，讓我們有更多的資訊可以選擇。

無論主流或是公民媒體，我基本上都抱持著一種「照單接收、但不信任」的態度；理論上（只是理論上），真正的事實來自不同資訊來源之間的交叉比對

現在，我們對於新聞來源有更多選擇、但也必須自己幫自己篩選。基本上，沒有任何來源是完全可信的；甚至有很多資訊來自不能說的弦外之音。如果不花那個力氣去幫自己判斷、比對、過濾新聞，那麼我們接收到的訊息品質，其實跟媒體開放之前沒有太多差異。

因为懒惰没有技术上的解决办法。

大多数人既然是人就继承了作为生物而应该天生拥有的进化本能和进化能力。长久的观察使我得出一个结论，大多数学生实际上是勤奋的，他们并不懒惰，他们只是花费了很多时间精力却没有收获而已。过分简单地把最终的收获归结于“勤奋”，而把颗粒无收归结于“懒惰”，只不过是于事无补的空洞评论而已，不为解决问题提供任何线索。

1) 不知道该干什么

We don't want to argue that noise is always good, it's clearly important to spend some time without it every day.

Filtering isn't everything it's cracked up to be, though, and you wouldn't want to live in a fully filtered world all the time. Social media noise is an essential part of learning and living on the web.

It's one thing to find something you didn't know you needed right now, it's a whole other skill to be able to recall information that seemed marginally useful at best in the past at a time in the future when the need for it arises. Who can't remember doing that before? The ability to recall passively collected information that was gathered purposelessly in the past and put it to use in the future is a particularly powerful form of intelligence

Doing well can mean several things: What you state is factually correct. Many people know or appreciate your work. The performance of your tool  or system is competitive with respect to some measure. You are getting a lot done. You are making a lot of money.

For most tasks we accomplish, no quantitative measure is satisfying

Most non-trivial work is not without errors. The best way to avoid errors is to do trivial work.