Notes

We want to help users, find great questions for increasing their knowledge and hence their reputations via their answers on the website stackoverflow. There are 1.5 million questions on stackoverflow and around 50000 tags on stackoverflow. Among those tags are some very important tags/topics like Operating system, data structures, algorithms, python etc. We want the user be able to understand such topics by answering questions them. So the more the user answers, the more will be his reputation via the score he gets on the answers and thus more will his knowledge. So mutexes vs semaphores can be thought of a single question which we want the user to master, we show him a question on it, and then show him linekd questions to him expecting him to answer the question. Only this way he will be able to learn the question and among all the questions on stackoverflow about mutexes and semaphores, he answers any question and achieves high scores means he was able to learn the concept. So the task would be to figure out all the questions in stackoverflow about a topic or multiple topics, figure out what questions he needs to master and then keep showing him linked questions to get him to answer them. We can use the score of the question to figure out how popular the question is, and also relatively how many users answers the question on it, and among all the tags present in those questions which tag is the most important. for example, given a tag how can you figure out the most important question.
so first step would be to first find out these concepts worth studying this is I already know. But find out the questions in these topics which have a high score which will tell a lot of about the importance of the question. maybe find out all the questions that have a high score and then make sure the user reads, understands and answers them. Many users just browse it to understand it a bit, but then forget it. We want to make sure they understand the concept as well answer one of the linked questions and then move on to similar questions.

How do you find out which questions are the ones that are the most important in terms votes.
Find all the questions in data-structures, find which question when asked led to the most increase in the reputation. We cant use the history of the user to find out anything about the important of the tag. But we can use to find out how much the reputation of the user increased on asking that question. If your answer receives an upvote, will give you 10 points. It is as simple as that. So we need to suggest questions most probable to receive answers from the user. but he wont get any score if just keeps reading the answers. so immedaitely after he clicks a question, we need to start suggesting him questions that require answers. But if he still isnt able to answer a question, then we need to suggest him anyone of the linked questions. But if he does receive a very good score, then we need to suggest him a similar question which covers a different concept. But then we need to find what is meant by a good answer.
Also how do you find the concepts to teach to the user? suppose the concept is operating system. How do you find out questions related to the operating system that are really hardcore and related very strongly to an operating system? I think this is done by the similarity search. Find tags are very near in similarity to the topic and then use them together to find out the questions important. What is the chance of getting an upvote? This is what we are concerned with. Which question if answered will lead to an upvote? How do u know if the question got upvoted and when? Find out at the what stage there is a higher probability of the question to get an upvote. We aren't interested in the question getting upvoted. But any of the answers. find out when an answer got an upvote, and what was the score of the question at that time. Just find at what stage of a question with a score if answered gives the user who answered a quick score. We don't him to answer a question which has been answered already but a new question waiting to be answered. Suppose we suggest a question to him that is similar to the above tag, but then still show him questions which will help him answer the question in a better way. And he can't simply copy a question because in such a way he will get a downvote. But then we can also deviate and suggest a question that is different from the current context, for example, python and mutex which wasn't there in the original question. But this type of questions is shown just to help the user solidify his knowledge as well have his reputation increased. Also a question which is old has a lesser chance to receive an upvote