1. If there are posts with comments, where each comment has several attributes (features). Design an algorithm which can determine whether the post is fake or not.
a. What do you do in case some comments have missing features?
b. The features are changed such that instead of a single value they have an array of values, what do you do to adapt it to the algorithm

Question

1.	If there are posts with comments, where each comment has several attributes (features). Design an algorithm which can determine whether the post is fake or not.
a.	What do you do in case some comments have missing features?
b.	The features are changed such that instead of a single value they have an array of values, what do you do to adapt it to the algorithm

Anonymous · Accepted Answer

Answer:
1.	SVM which gets the features and separates between fake or not fake. There needs to be an explanation which deals with soft SVM (in case the classes are not separable) and hard SVM.
a.	Either remove the values (if there are only a few of them) or complete them by average.
b.	Calculate several (few) features from the array of features. Again, by average, std, median and so on. It’s also possible to calculate new features using PCA.

Microsoft

Microsoft Interview Question

Interview Answer

Want the inside scoop on your own company?

Bowls

Followed companies

Job searches