Data Miner, Data Analyst, software developer
About Me
In my research work at university of London, I proposed a categorization based search engine and evaluated it in the experiments of simulated setting and real user data setting. - I implemented the whole experiments involving data acquisition, data analysis,
results visualization and interpretation. The results confirmed that my approach improved users search behaviour by 22% in terms of target document rank.
I carried out a topic analysis of 15 million search queries from Microsoft Live
Search Engine - I built a predict model to categorize the queries and evaluated the performance, the model yielded 75% on Macro-F1(Accuracy). My results were used to understand the statistical characteristic of the user queries based on the topic.
I
designed a fast learning algorithm on a large matrix decomposition - I conducted the experiments and derived an error bound on the algorithm.
I Implemented a neural network model for learning to rank category.
