Worked on the problem of improving the content quality by filtering NSFW contents on the platform.

Process Mining refers to extracting useful information about the processes from large history logs. In the Process Discovery, for a large log a very complex model can be formed which would be very difficult to comprehend and make the discovered model worthless. So, a important part from business perspective is to find abstractions or patterns in the log and abstract the model according to that in order to get overview of the log. There are several methods to get pattern but they take into account only the timeline of the process neglecting the other information of the log such as resources, department, etc. These methods are basically based on frequency model therefore, it gives a lot of wrong patterns(false positive). My work here is to come up with some way to prune these patterns with the help of other attributes(resources) of the log using data mining techniques and machine learning. • Aimed at achieving better extraction of useful data from big data logs using data mining techniques. This work aimed to improve understanding of people on their businesses by finding the bottleneck.• Proposed a method of using resources rather than just timestamp in the log for getting better abstraction• Proposed a prefix matching heuristic for resource clustering. Implemented the same and found promising results• Proposed a method to use decision tree to come up with the number of clusters into which resources and activities need to be partitioned. Implemented the proposed methods and found a considerable rise is accuracy from 46% to 86%

• A Microsoft SQL based Business Intelligence project with different reports to gain insights and better business planning• Created the frontend reports which systematically provide all the data required by the user • Created Stored Procedures to create SQL tables, did the code quality check, verified the reports for the UI guidelines