The master thesis utilizes machine learning clustering algorithms to divide malware into distinct groups and evaluate the performance with V-measure. The procedure is as follows: 1. Collect malware of different families from a convincing source 2. Use Cuckoo sandbox to dynamically analyze malware samples 3. Eliminate inactive samples provided with some thresholds 4. Transform API Statistics and API Sequences of Cuckoo analysis reports into vectors 5. Include Feature Selection to retrieve valuable features and discard less important ones 6. Employ clustering algorithms to separate malicious samples (KMeans, MeanShift and BIRCH) 7. Compute V-measure scores to measure efficacy

The Final of Taiwan College Campus Software Design Contest (https://goo.gl/ty0fxh)－2009 A SOCIAL Network system: show relationships between professors by using Web Crawlers to fetch data from Internet (Java), extracting useful information and storing it in MySQL Database