| | | |
| | |
| |
| | |
| |
| | |
| |
The microblogging service Twitter allows users to broadcast short messages to their followers. Users can express their opinion, describe experiences, and spread ideas and information using 140 character limit. Recently the number of twitter
users has reached to a 75 million and the resulting flood of data can potentially be mined to discover the buzz about products, people and events. We propose semi-supervised Labeled Hierarchical LDA and demonstrate lhLDA improved expressiveness by merging related terms into one topic and outperformed over hLDA by clustering similar topics under the same parent.
| |
| |
Download: PDF
| |
| |
| | |
| |
Implement generative model approach for unsupervised POS tagging using EM training.
| |
| |
Download: PDF
| |
| |
| | |
| |
Define a knowledge representation scheme and create instance frames representing sentences and build a sentence generator and implement multi-sentence text processing
| |
| |
Download: PDF
| |
| |
| | |
| |
Implement CKY Parser given a probabilistic context-free grammar
| |
| |
Implement Knuth 77 algorithm (best-first dynamic programming algorithm)
| |
| |
Implement an Earley parser (left-to-right parsing with top-down filtering via prediction step)
| |
| |
Download: PDF
| |
| |
| | |
| |
Design and Implementation Berkeley DB cache manager
| |
| |
Implement Berkeley DB DRAM and Magnetic Disk
| |
| |
Based on Greedy Dual Cache Management Techniques for Mobile Devices
| |
| |
Download: PDF
| |
| |
| | |
| |
Generating Ontology from Flickr Tags
| |
| |
Mine a corpus of tagged photos from Flickr
| |
| |
Input: location name or address / output: location based information visualization in tree and triples
| |
| |
Download: PDF
| |
| |
| | |
| |
Mine a corpus of tagged documents from the social bookmarking site del.icio.us
| |
| |
Source discovery: finding similar web-sites with seed source based on the tags, users, documents
| |
| |
Learn latent topics using LDA and gather related sources using JS divergence
| |
| |
Download: PDF
| |
| |
| | |
| |
Matlab Neural Network Toolbox to create neural networks recognizing faces and motions of objects
| |
| |
Design and train a neural network to distinguish 20 different faces regardless of their facial expressions
| |
| |
Design and train a neural network to recognize each of 5 different motions (Up, Down, Left, Right, and
Diagonal)
| |
| |
Download: PDF
| |
| |
| | |
| |
Implement Searching algorithm in grids with blocked and unblocked cells: A* and Theta*
| |
| |
Download: PDF
| |
| |
| | |
| |
Design and implement expert system that performs inferences the probability of winning in Monopoly
| |
| |
Download: PDF
| |
| |
| |
