Academic Advisor: Dr' Yuval Elovici

1
ADD Presentation

• Academic Advisor Dr. Yuval Elovici
• Technical Advisor Dr. Lidror Troyansky

2
Where do we fit in?
Continued
3
What are we doing??

• Develop a system which will
• be able to Configure the searching parameters.
• scan the P2P networks.
• download files suspicious as confidential.
• analyze the material using Machine Learning.
• generate reports.
• produce statistics.

4
Requirements
5
Functional Requirements

• Scanning and looking for suspicious target (e.g.
  as confidential) information in the P2P network
  (Gnutella).
• Downloading the suspicious target (e.g. as
  confidential) information from the P2P network
  (Gnutella).
• Analyzing the scanned results (determine the
  value of the documents).
• The system will use the Machine Learning based on
  the filtering algorithm to classify the
  documents.
• Statistics Gathering
• The number of users which currently hold the
  target information.
• Using IP Geolocation and finding out the
  geographic location of the leaked information.
• The history of searched for, downloaded
  analyzed files.

6
Non-Functional Requirements

• Performance constraints
• The system should return a search result for
  suspicious target after no more than 15 minutes.
• The system should not limit the download target
  time. (Remark it should be configurable. By
  default, a time-out should always be set)
• The system should hold history result and
  statistics of not more than one year ago.
• Safety and Security
• The system will not be used for any other purpose
  than find information leaks in P2P networks (e.g.
  to find MP3 shares).
• The system will not expose the confidential
  documents it downloads and the documents were
  used in the Machine Learning algorithm.

7
System Architecture
8
System Architecture (cont'd)

• The system is constructed from several components
  which are written in different languages and
  communicate between each other in several ways.
• All software modules reside in the same computer.
• IGTellaHandler- The primary responsibility of
  this component is downloading documents from the
  Gnutella Network. The IGTellaHandler is written
  in Java and communicates with the main component
  (P2PinspectorGadget) via RMI technology (to
  increase the de-coupling between the copmonents).
  
• IGConfClassifier-The primary responsibility of
  this component is classifying documents using
  different classification rules.The output of
  this process will be saved in the database, and
  will be available for further use.

9
System Architecture (cont'd)

• IGDBHandler- The primary responsibility of this
  component is connecting to an external database
  and stand as an interface for the system's
  modules for the database.IGDBHandler will be
  written in java and will communicate with the
  main component via RMI communication.
• P2PInspectorGadget-This component is the
  system's main component, it has two primary
  reponsibilities, the first is interaction with
  the user via the Graphical User Interface, and
  the second is to control the flow of the
  system.P2PinspectorGadget will be written in
  Java and will connect to the different components
  with the connection mentioned above, and will not
  communicate with any other external system.

10
Main Classes
11
Main Classes (cont'd) Network handler
12
Main Classes (cont'd) Network handler
13
Searching files seq. diagram
14
Testing

• Unit testingAll the units will be tested for
  every use case.For each use-case all of the
  possible paths will be tested.The unit testing
  is a part of the design of the project, an
  automated tests are running all of the time when
  we develop the system.Here are some of the
  testing in the test-plan
• Start system Starting the system with a
  firewall blocking of the P2P needed ports, and
  see that the system doesn't crush and outputs the
  right error message.
• Scan network Verify that this process concludes
  after a pre-defined time-out.
• Analyze downloaded files Verify that the system
  converts the different text formats (DOC, PPT and
  PDF) correctly into "raw" text.
• Analyze downloaded files Verify accuracy of the
  algorithm (achieving the standard of
  false-positive and true-positive as defined in
  the project's targets.

15
Testing (cont'd)

• Acceptance Tests As a part of the acceptance
  tests, all of the use cases will be fully checked
  from the beginning to the end.In addition, all
  of the non-functional requirements will be tested
  to make sure they meet their targets
• System's HistoryIn order to verify that the
  System saves all the information for the period
  that the user has defined (default is 1 year), we
  shall manually try to change the system's clock
  and trick and see that the data that needs to be
  saved is saved and the data that should have been
  deleted, is deleted.
• System legitimacy (non pirate uses)The system
  will be blocked for uploading data, this will be
  checked with planting a unique media file (maybe
  MP3, or MPEG) that we composed, with a unique
  name, and try with a different client to download
  the media file.

16
Testing (cont'd)

• Content SafetyIn order to test for Content
  safety (classified documents used for the
  learning part of the algorithm will not be
  exposed to the P2P network), those two
  sub-application are running as a separate
  processes with different memory space.The test
  will be attempt from another client to download
  the classified documents or the list of the
  documents from the process that connnects to the
  P2P network.

17
GUI
18
GUI (cont'd)
19
Remaining tasks (till prototype)

• Create and Integrate the GUI.
• Find a list of Gnutella1 working servers.
• Classification algorithm inspecting and learning.
• Integrate Python written algorithm to Java.
• Finish PDF 2 DOC converter.
• Finish Gnutella driver (able to perform search
  and download of documents).