What topic to choose for a master's degree?

Question

data mining

Hello.
I am trying to choose the direction and approximate topic of my future master's thesis.
Bachelor's work was on the topic: "SaaS DSS" (in fact, an interactive decision tree for the SaaS model). Plus the second diploma, which I didn’t do for myself: determining the subject of the text (implemented a naive Bayes classifier with the selection of key features of the text according to the chi-square criterion).
I know java, android, javascript well , I can build relatively simple web applications. I did some visualization of algorithms and data. Like algorithms on graphs/networks. The tasks of data mining and information retrieval are interesting . Now I'm reading the book Introduction to Information Retrieval - I like it .
Recently, an interesting, as for me, task of recognizing emotions on a face appeared in my head. But firstly, there are already quite a lot of ready-made solutions ( habrahabr.ru/post/133686/), including a web solution, and secondly, it seems to me that I can not pull it ...
Clustering algorithms are also interesting, but for this There are so many things already on this topic that I'm not sure if this will be relevant.
From the master's thesis is required:
1) novelty, or if such projects exist, then my project must have some advantages over existing ones
2) scientific research.
In general, the requirements roughly correspond to this article habrahabr.ru/company/spbau/blog/220267

Answer 1

Data mining is a very interesting topic. Russia has a fairly large set of open data (and more ), which are just asking for analysis. Try to find interesting options with them.
For example, I liked the story Data mining helped earn the Pulitzer Prize . There was also a very practical article Data Mining in online games (and the second one ), which gave amazing results. Or here is the article Trading knows when you're expecting a baby , which led to a scandal.
In general, Data mining is a very promising topic, and in Russia it has not yet been thoroughly dug, it will be possible to gain material and novelty not only for a master's, but also for a PhD thesis.