"Web Based Mining and 3D Rendering of Knowledge Structure"

Views:
 
Category: Education
     
 

Presentation Description

Computer Science Spring 2013 Student Research Presentation by Stephen Cohen

Comments

Presentation Transcript

Web-Based Mining and 3-D Rendering of Knowledge Structure:

Web-Based Mining and 3-D Rendering of Knowledge Structure Stephen M. Cohen May 2013 Graduate Capstone Project Presented by: Stephen M. Cohen Advisor: Dr. Xiang Fu May 2013

Outline:

Motivation Design Web Based Mining Visualization Experimental Results Outline

Introduction: Adaptive Testing:

Introduction: Adaptive Testing Write a program for primality test ? ALOAS http://people.hofstra.edu/Xiang_Fu/XiangFu/Projects/ALOAS/index.html Loop Branch Integer division

Ontology:

Ontology [1] Behind Testing Tool Manual Construction Too Tedious ! Ontology Example of an ontology(topic map)

Purpose of Research:

Given Scientific Subject Extract Nodes of Ontology Automatically Our Contributions Purpose of Research Web-based Mining Tool Visualization Tool Evaluation and Experiments

Facts of Project:

Windows Form App 9627 Lines of C# Code Bonus Feature: Google Drive Backup XNA 4.0 Web Search Tree Visualizer Generates XML Facts of Project

Part I. Web Based Mining:

Part I. Web Based Mining

Web Mining Machine:

Web Mining Machine Input Start Link (EN Wiki) Recursion Depth Max Links/Terms Per Page Output

Algorithm:

Essentially Recursive Algorithm Parse Page Using Regular Expression Special Wiki Links  Terms Precision Control Based on Distance to Root/Target Term Extra User Provided Oracle : List of Core Terms Algorithm

Screenshots of the web crawler:

Screenshots of the web crawler Before After

Part II. 3D Visualization:

Part II. 3D Visualization

Algorithm:

Generates Balanced Search Tree Labels Nodes With URLs Labels Paired Nodes Connects All Nodes With Red Lines Algorithm

3d Web Search Tree Visualization Tool:

Input 3d Web Search Tree Visualization Tool

Visualizer Angled Camera Screenshot:

Visualizer A ngled C amera S creenshot

Zoomed Out:

Zoomed Out

Node Highlighting:

Node Highlighting

Node-by-Node Navigation:

Node-by-Node Navigation

Display Optimization:

Display Optimization

Features:

Features Navigation Controls http://lnk.nu/youtube.com/2c1t Standard Controls Plus-zoom in Minus-zoom out N-look up M-look down V-look left B-look right Q-Quit Left Shift-optimizer switch Special Viewing Mode Controls Space bar=center camera on top node of tree(or the last node you centered the camera on using up, down, left, or right) Up-go up to parent node Down-go down to child node Left-go left to the adjacent child node sharing the same parent node, or to the last child node sharing the same parent node if you are at the left side of the child nodes. Right-go right to the adjacent child node sharing the same parent node, or to the last child node sharing the same parent node if you are at the right side of the child nodes.

Part III: Evaluation:

Part III: Evaluation

Sample Pairs Generated:

Sample Pairs Generated

Problems and Solutions:

Data Structure Construction Fast ,Rendering Speed Slow Optimization Switch Added Performance decreases as Search Tree Size increases because of increased rendering times Problems and Solutions

Results:

Discrete Mathematics 4 links per page depth of 4 - 86.50% accuracy Discrete Mathematics 3 links per page depth of 5 - 85 % accuracy Computer Science 5 links per page depth of 5 - 78 % accuracy Results

Bibliography:

1. Lin, Feiyu. “State of the Art: Automatic Ontology Mapping”. PDF Document. Information. Engineering Research Group .Department of Electronic and Computer Engineering School of Engineering, J¨onk¨oping university J¨onk¨oping, Sweden. January 2007. Bibliography