Supervised Categorization of JavaScript using Program Analysis Features
Paper
This is the homepage for the paper titled "Supervised Categorization of JavaScript using Program Analysis Features", which appeared in the 2nd Asian Information Retrieval Symposium (AIRS 05). A copy of the paper can be downloaded from here.
Presentation
Presentation slides can be found here.
Dataset
Our dataset is derived from the standard WT10G corpora (More details on the WT10G corpora can be found on the TREC website and at the CSIRO website Anyone wishing to use this corpus must sign an individual license agreement before proceeding).
A complete dataset containing 18,683 documents can be downloaded from here. You can also download these documents grouped by servers from here.
We have selected 1,637 instances (also considered as "JavaScript units") from these documents for experiments. This is the list of the units. All the JavaScript units are constructed dynamically by our system. The list file looks something like this:
SERVER1168_WTX098-B36-310_N_0_0$form_validation $
SERVER1168_WTX098-B36-310_N_0_0$form_validation
$
This line means the following: this JavaScript unit is extracted from a document
named "WTX098-B36-310" in the WT10G corpora,
and it is associated with a server of ID "SERVER1168".
It is an interactive unit ("N": interactive
unit, "D": non-interactive unit), with unit
ID "0" (the next digit "0" is for
verification purpose). This JavaScript unit is annotated as a "form
validation" class.
System
You may construct JavaScript units using our system. Our system is written in Java and is under GNU license. You can download the source code from here (updated 7th July). You should re-compile the system by yourself. Sample files for utilizing this system can also be found in the source package.
The system contain partial modified codes from the Rhino 1.6 and HtmlUnit 1.4 (Please read their license agreement before proceeding). Modified codes are included in the system and you should NOT download the original codes. Some external libraries are also required. You may also find weka helpful in doing evaluation tasks.
The system was developed and tested with J2SDK 1.50.
Required external libraries (this list may be incomplete):
| jaxen-1.0-FCS-full.jar | XPath support |
| saxpath-1.0-FCS.jar | XPath support |
| commons-collections-3.1.jar | Collection classes |
| commons-lang-2.0.jar | Core Language Utilities |
| commons-httpclient-3.0-rc2.jar | Provides the actual http support |
| commons-codec-1.3.jar | Provides common encoders |
| xercesImpl-2.6.2.jar | XML Parser |
| xmlParserAPIs-2.2.1.jar | XML Parser |
| nekohtml-0.9.4.jar | Converts html into an XML DOM model |
| commons-logging-1.0.4.jar | Logging support |
| commons-io-1.0.jar | IO utilities |
This page is maintained by LU Wei (Mr.) Last updated: Thu, July 07, 2005, 00:50:17