Xin Ye

We developed a ranking model to recommend source code files to fix for a bug report.
The ranking model is a weighted combination of multiple features that every feature is a unique type of information measuring the relevance between a source file and the bug report.
The weight parameters of our ranking model is trained automatically using a learning-to-rank technique.
Through evaluations on six large-scale open source Java projects, our model achieved significantly better results than the state-of-the-art approaches.
Languages used in this project: Java, Python, SQL
Tools/Libraries used in this project: Git, Bugzilla, MySQL, JDT ASTParser, Apache Tika, Beautiful Soup, NLTK, JUNG, SVM^rank
Algorithms used in this project: VSM, SVM^rank, PageRank, HITS
[Paper] [Slides]

We developed a 3-D virtual boat (VB) game running on iPad devices for a NSF GK-12 project to assist high school teaching in Ohio.
Our VB game contains multiple scenarios in which each scenario corresponds to a dedicated lesson plan designed by a NSF GK-12 fellow.
Our VB game and the corresponding lesson plans have been using in the classrooms of 9 high schools in Ohio.
Languages used in this project: Javascript, C#
Software used in this project: Unity, Xcode, Google SketchUp, Autodesk Maya
[Website] [Youtube]

We developed a 3-D mobile chemical experiment game for high school education.
The game mimics the classic Winkler Method that was used to detect dissolved oxygen in water.
Languages used in this project: Javascript
Software used in this project: Unity, Xcode, Google SketchUp

We developed an interactive 3-D visualization tool for visualizing software projects’ version control repository.
In the visualization, a source code file is represented as a set of cylinders in which a small cylinder refers to a specific revision of this file.
Different colours of the cylinders refer to different developers/authors who contributed to the specific revisions.
Languages used in this project: Javascript, php, C#
Software/Tools used in this project: Unity, Eclipse, SVN
[Paper] [Youtube]

I developed a search engine for searching websites within the ohio.edu domain and other websites related to Ohio University (OU).
I used Apache Nutch to crawl 117,620 web pages based on 10 OU websites as the seeds.
I used Apache Solr to index these 117,620 pages.
I built a search engine as a servlet running under Apache Tomcat. The servlet uses Apache Lucene libraries to search web pages within the index created by Solr.
This search engine supports VSM, BM25, and the Smoothed Unigram Language Model. It also uses anchor text for searching.
Languages used in this project: Java, HTML
Tools/Libraries used in this project: Apache Nutch, Solr, Lucene, Tomcat, Servlet
Algorithms used in this project: VSM, BM25, Smoothed Unigram Language Model
[Demo] [Report]

I developed a face detection program to automatically detect human faces from static images.
This program applies Principle Component Analysis (PCA) to create an eigenface subspace from the training images.
A training or a testing image can be represented as a weighted combination of the eigenfaces.
Therefore, an image can be represented by its weight vector in a lower dimension.
If the euclidean distance between the weight vector of a testing image and the weight vector of a human face image is below a threshold, then face detected.
Languages used in this project: C++
Tools/Libraries used in this project: OpenCV (compiled on Solaris)
Algorithms used in this project: PCA
[Report]

We developed an iOS stock exchange application that can be launched on iPhone and iPad.
Users buy and sell stocks with virtual cash to compete for the most profitable day trader in the app.
The app fullfiled the user requirements from 5 wide, LLC in Cincinnati.
The app was developed by a small team with 4 developers that used SVN for version control and Redmine for issue tracking.
Languages used in this project: Object-C (font-end), PHP (back-end), HTML (back-end), SQL (back-end)
Tools/Libraries used in this project: Xcode, iOS 4.3 SDK, MySQL, Apache Webserver
Version Control: Subversion
Issue Tracking: Redmine
Memberships: iOS Developer University Program – Ohio University (Electrical Engineering and Computer Science)
[Youtube] [Report]

I developed an IPv6 router.
It responses to ICMP Echo request, UDP Echo, ND request, Router Advertisement, RIPng advertisement, ICMP error message, time exceed, and udp port unreachable.
It supports ARP table, routing table, and routing queues.
I used both the Times 33 and the Times 31 algorithm for double hashing to avoid hash collisions.
I used multi-threads in this project.
Languages used in this project: C++
Libraries used in this project: POSIX Threads

I modified and ran U-Boot on IBM PowerPC that was integrated within the Xilinx Virtex-II Pro FPGA on a digital signal processing board.
For debugging, I used U-Boot to load MontaVista Linux kernel and ramdisk image through TFTP.
For stable release, I used U-Boot to burn itself, the MontaVista Linux kernel, and the JFFS2 filesystem to a NAND flash device on the board.
Languages used in this project: C
Tools used in this project: U-Boot, MontaVista Linux