Home : Projects : Resume : Contact : About

The RAE Biomedical Information System <Closed Source> <Midterm Progress Report>


Description <Click Text To Expand>

RAE Desc
    The RAE Biomedical Information System is a partnership between Dr. Mark Sangimino, an Orthopedic Pediatric Surgeon at Allegheny General Hospital, and Dr. Mandy Raab, Director of Biomedical Informatics at St. Vincent College. Development is assisted by Dr. Cynthia Martincic, Associate Professor of Computing and Information Science also of St. Vincent College.


Large Document Search Engine <Project> <SVN> <Download>

Current Change Log <Click Title To Expand>


Description <Click Text To Expand>

LDOC Desc
2010 Senior Project

    As a purely scientific endeavor inspired by the work by Dr. Patrick Juola on JGAAP which deals with authorship attribution. This project in contrast attempts to identify contextual similarities in document content by analysing the commonality of regular and proper noun frequency on the sentence, paragraph, and whole document levels. Which this information it is possible to compare one or more documents for contextual similarities with the hope of allowing a user to essentially ask a single author a question or comparing the responses of multiple authors to a topic. This is an open source project which hopes to inspire the further study of large document contextual analysis.

Purpose:
    1. To define an algorithm that will accurately define the components of a document which can be related to other documents to show similarities between the context of the documents.
    2. Provide a system to analysis and categorization of document archive to permit users to request specific comparisons between bodies of work. ie. To quickly discover bodies of text within different supplied documents which relate to a common theme.
    
Process:
    1. The algorithm defined to perform this task will be strongly dictionary based. First document items will be categorized grammatically permitting the extraction of key items without parsing out special characters and markup. Items that do not follow standard dictionary identification will be processed for special characters and then identified as keywords. Generally these items that fall to the bottom of this process should be proper nouns which will act as context keywords. In conjunction with the proper nouns the system will also create so called, "Lower Order Words" (LOWords), which will be defined as the comprehension words for a single word. This is accomplished by processing all nouns through a thesaurus. The list returned from the thesaurus lookup will then be weighted by taking each item from the first list and performing a thesaurus lookup on each item. This final large list of LOWords will then be processed to only return the top 5 most common items. These items are then stored with each noun.
    The document will be broken down to the word level with subdivision for sentences and paragraphs. These LOWords will be most abundant at the word level and less abundant at the document level. For example each sentence will contain the top LOWords from all the words that sentence contains and each paragraph will contain the top LOWords from all the sentences in that paragraph.
    2. Analysis will be done by comparing paragraphs of documents LOWords and keywords to each other. These percentage of matching will then be returned to the user.

Product:
    This system will be constructed in two forms. Primarily and initially, this engine will work in conjunction with a GUI to permit local desktop access to processing and analysis. This analysis will be between documents and documents and user queries. Time permitting this system will be able to be run in a server environment to provide a web interface for document analysis and storage of a processed document library. Its function will be similar to that of the desktop client but will provide processed document persistence and passive background processing of document stores like Project Gutenberg and JSTOR.



Auto Kindle eBook Converter <Project> <CVS> <Download>

Current Change Log <Click Title To Expand>


Description <Click Text To Expand>

Auto Kindle Description
Project Outline: Create a simple converter using command line tools that will detect filetypes and convert documents in .lit, .chm, .pdf, .pdb, .html to .mobi for use with the Amazon Kindle

The best way to look at this project is to simplify the conversion process. There are definitely more mature applications that do the same tasks e.g. http://calibre.kovidgoyal.net/
This project may end up in that direction in the future but I live in a world of portable data. I want to just be able to grab an exe and run the app. Many of these larger projects require installation and can be bloated. Not that they are not fantastic for the low lever computer user but any time you combine your tools into one large app there is the possibility of the application losing focus. My best example is iTunes. Not that I am going to make the statement that calibre is going to go that way. Of course not it is an open source project and we all know that open source works so I have faith. My world is that if I were to make a library to manage a users eBooks along with my converter. All it would do is call upon my converter to do the work. Not one large app but many small apps. Actually this is the direction used in calibre and I applaud it. But his small app mentality starts and stops at the converter and device driver level. This could go either way.

This project was originally a quick and dirty method for me to convert my ebook collection in various formats to a kindle compatible format without worrying about the ebook source type. It takes PDF, Lit, PDB, CHM, and HTML files and converts them to .mobi.

Changes in this version. Major bug fix with filename rendering and added the ability to change the title metadata for mobi files and forces the filename as the title of PDB files which in the past would show no title or gibberish. Added an option to write file in place by canceling the destination dialog. No info is in the app about this. Just a place holder for convienence.

To take advantage of Drag and drop feature right click on the icon in your start menu named Auto Kindle Ebook Converter and select Send To ---> Destop(Create Shortcut) Now any time you wish to convert a book just drag the book onto this icon and tell it where to save the new file. Everything else will be handled for you. Enjoy



Universal Character Sheet System <Project> <CVS> <Download>

Current Change Log <Click Title To Expand>


Description <Click Text To Expand>

Universal Character Sheet System
Paul Scarrone 12-22-2008
Project rewrite. System will not be single page website utilizing JS and no XML.
    The misconception I had was that XML was needed to store the page information. The truth is that an HTML page is a XML file through the DOM. In this same way the page can be self saving. By saving the changed DOM and a page template the page can be made to have the impression of saving itself. In reality saving a new copy will be the standard until I can work out any bugs with the system as I have experience a full page wipe using the current objects in a small scale test. Obviously this is not a huge deal until release.

I have rebranded this under the RPHaven project of old at sourceforge. Initial files are currently in the CVS for the project. I will be continuously working on these as a multipart website until I have a working enough copy to attempt integration into a single file website.

Changes include auto layout with DIV containing field information and edit and delete options. All option dialogues will be model and exist in the same page for easier passing of DOM information.
For future editability all popup option boxes will be stored in <textarea> fields at the base of the page. via CSS these will be not visible unless page is in edit-mode at which point thew will display and allow users to change the popup dialogues themselves to incorporate new features for new systems.

The data for a field will be store in the field itself and be referenced via hover even or alt for image. fields will be copyable and editable.
CSS education will provide some plesant visuals.

Current data per field will consist of The title of the field if needed unless field is in a parent field the data type. a Description of the field this can involve rules to be displayed to user. and tenative is a rule logic creator that will generate a string operation to affect a given amount like an excel function. Layouts in this current version will be static But I plan to incorporate some kind of way to move the fields around.

Consider JQuery for JS library for app
Paul Scarrone 12-22-2008

Paul Scarrone 10-7-2008
TODO:
Explore self contained website solutions MHTML ideals   (Completed 10-7-2008)
Explore self modifying static XML in this file without any PHP required (Completed 10-7-2008)
Explore self contained browser only applications via JS, HTML, XML  (Re-mapped) 10-7-2008
Expore Javascript Parser (handmade) to reference the local file, non IE xmlHTTPRequest
Paul Scarrone 10-7-2008

Paul Scarrone 11-6-2008
Explore Javascript Parse that will allow for dynamic creation of div tags and a system to create the content of those div tags and saving the file(11-6-2008)
Paul Scarrone 11-6-2008


Connection to Web-app tabletop system. Utilization of UCS will allow players to take characters from online game to offline game. allowing offline local sheet updates and then online synchronization.

Common language Definition.
The maximum possible types of data information about a character that could be needed. from static items to user definable items. Common language gives the sheets a meaningful representation to players and creators alike. Easy fix for complicated custom language is to add element mapping for full custom systems user can define location and effect. giving each element a custom quality. Requires second piece that the online tabletop system allows for. Interpretation will be done by human eyes or by users creating the correct connections within the system themselves. Even from a local standpoint as long as the sheet can display itself locally and perform calculations based upon basic criteria all is good. More detailed example to come based upon shadowrun character sheet.



Paul Scarrone 5-13-2008
XML markup for a universal character sheet system
need web based and client based editor.
 
Game types defined by schemas ( user uploadable ) Need A schema creation system to accompany user uploaded xml documents
 
small size and ability for small enclosures and a packaging system for aditional media like dialect and bg music and pics
 
re-launch universal character sheet system using XML and to create and a reader. Needs to have appropriate XSL system to allow for self display of any appropriate information
 
no binary file access
 
facilitate packageing of enclosures and still web loadable file without extraction.
 
Or need be for sheet and href enclosures possible use of mhtml file type conversion to mxml
 
 
Sheet will contain the rule-set for the game built-in to the file for custom rule-sets.
 
Ruleset for both character creation in xsl modified template mode.
 
xml and xsl will be self contained to permit a complete character sheet giving access to creation of more character sheets of the same type. without access to ruleset export.
 
Rulesets can be exported for dice rolling examples and game notes Including pdf documents of complete rules through pdf or word doc. for both existing games and custom ones.
 
game rule sets can be revisioned and I would like to work a google doc integration into the system or infile editing
 
CS check-in and check-out system for local remote non internet play but tracking of experience and changes.
 
game creators can re-upload 1 file
 
Interactions will be automatic incorporate google gears for offline content and data store. When character sheet is opened and user is online client will auto update rules associated with files and custom game descriptions. revisions will be maintained serverside and will be accessable online.
 

Paul Scarrone 5-13-2008 
Field Tags and properties of the tags. Universal language for generic xhtml generation
 
Almost an aspx conversion framework for creating the site elements.
 
Auto Save feature similar to google. Consider using google api for data entry


Adaptation of character sheet and ruleset system as a modular cartridge for a game engine. Even ability to load multiple carts at any given time for multiple effects. Similar to mutators for the unreal engine.
 
Rulesets and character rules can be merged by category and modified by override one set over the other.
 
Effective micromanagment of ruleset priority can be saved into modified cs rulesets and transportable

Paul Scarrone 10-7-2008
Some Design Issues
XML is ultimately the best design component for the data elements in this project but more importantly is being able to combine all data sources in the same file. Questions are can I modify an XML static table in an html file via javascript without throwing up huge security errors to the system. This self injection on demand could allow for viral injection into the code. There must be a safe way to execute this


SVC Driving Simulator <Project> <Download> <SVN> <Documentation> <Progress Reports>


Description <Click Text To Expand>

Driving Simulator
This is a project between the Computer Science and the Psychology departments at St. Vincent College via Dr. Cynthia Martincic and Dr. Mark Rivardo respectively. The purpose is to create a driving simulator suitable for testing human reaction.

    Initially the SVC Driving Simulator was an individual project assigned to me for educational purposes but later it was converted into a group project for Fall 2009 CS 355 Software Engineering class taught by Dr. Cynthia Martincic. The team consisted of six members of which I was the project lead and manager.
Team:
Sarah Anderson
Josh Woods
Nathan Hoffer
Riley Hardin
Anthony Williams
Paul Scarrone (Lead)
    This course covered the processes of agile development and iteration design. These techniques were applied to the project through the course of the semester with a successful completion resulting in a testable prototype. The prototypes focus was to prove the viability of creating an accurate rendition of a vehicle traveling on a road surface. This was achieved by rendering the road in realtime from complex vector which defined the roads center and then tracking the vehicle object in relation to the road surface. DarkGDK was utilized as a 3D environment for the project. DarkGDK is a simple wrapper for DirectX which handled collision and object tracking for the vehicle.
    Major hurdles in this project were the creation of irregular polygons to define the road surface and identifying if the vehicle was in fact on the road surface. DarkGDK only supported the creation of regular polygons and rectangles. To create an irregular polygon the triangles which made up the object would have to be drawn manually. To permit the sequential drawing of irregular polygons a stable draw order had to be developed that could stack one polygon upon another without creating artifacts on-screen. Once this data was calculated it was written to a binary object and stored to reduce startup time when using the same road surface and to permit the quick manipulation of the whole road as a single object. To verify that the vehicle was following the road surface an algorithm was developed that would find the road segment that was closest to the vehicle and then calculate the distance from the vehicle to the the closest point normal to the road segment.

Goals for the project:
Track Vehicle Location
Track Vehicle Angle
Track Speed/Deceleration
Track Side of Road
Track On or Off Road
Display this Information Graphically



This Website <Project> <Mirror> <Project Documentation>


Description <Click Text To Expand>

Untitled Document

There is a 2 fold plan I had when building this website on my schools server.


First I wanted to have a "valid" repository for the work I am doing here.

Second I wanted to be able to express myself about my school life from within the school.


Of course many of the views and projects I am involved in are not neccessairly associate with St. Vincent College but they are all part of my education that I feel this College cares dearly about. In that way I am giving something back by discussing my positive experiences that happen to me on campus. In the hopes it will give a deeper example of what life is like here for people like me to attend. Candid exploration of this instutution has only allowed me to fall in love with it further. This is my method of expressing that love.

 

On the technical side this website is a little on the lighter end of the things I have been working on. It uses basic PHP to load documents inline from google documents. As I personally find it easier to format and work with text documents within an actual editor. This will be the only text involved in this site that does not load from offsite. Although even this section is inlined from another local webpage thissite.html.

Outside of this I created a JS class to handle the expandable div sections I always find JS more fun but this is not a strong example of my skills but it was fun none the less. I do believe that thinking of all operations as objects is a strong way to keep your mind programatically sharp even though there was no need to build these functions into a class.