HexTAC: Human EXtraction for TAC
Description
HexTAC (Human EXtraction for TAC) has been developped by the RALI laboratory of the Université de Montréal for the TAC 2009 Workshop. Its purpose is to simplify the creation of extractive summaries. It is a browser-based interface which enables a user to build a summary step by step in a convenient environment.
HexTAC's interface is generated by an XSL stylesheet which takes a source text in XML format as input. The Javascript code (client-side) of HexTAC allows a user to manually produce an extract of the document and to send an XML version of the extract back to the server. HexTAC can also be used to modify an extract that have been generated by the interface. Such an extract has to be provided as a paramter of the XSL stylesheet.
Here we provide the client-side part of HexTAC as well as 2 minimal PHP scripts that run the server-side part. These scripts use an example XML source document and an extract of this document produced by HexTAC.
You can use HexTAC to produce your own manual extracts, provided you develop the adequate server-side application.
HexTAC is distributed under the GPL licence version 3.
For further details on HexTAC motivations and experimentations please refer to the following articles:
- Genest P.-É., Lapalme G., Yousfi-Monod M., HEXTAC: the Creation of a Manual Extractive Run, in proceedings of Text Analysis Conference (TAC) 2009, Gaithersburg, Maryland, USA, November, 2009 (PDF)
- Genest P.-É., Lapalme G., Yousfi-Monod M., Jusqu’où peut-on aller avec des méthodes par extraction pour la rédaction de résumés, in Proceedings of TALN 2010, Montréal, Québec, Canada, Jully, 2010
Try HexTAC
You can try HexTAC on the source documents and their corresponding summaries on this demo page.
Download
If you want to install HexTAC on your own server, you can can get all required files with the following package:
Content
The following describes the purpose of each file of HexTAC.
| HexTAC_generate_interface.php | A server-side PHP script that applies the XSL stylesheet to the XML source document to generate HexTAC's HTML interface. If an extract filename is specified as a parameter, it will be loaded and ready to be modified. |
| HexTAC_display_extract.php | A server-side PHP script example called by the interface's HTML form. It recieves the new extract and other informations and display them. |
| HexTAC_doc_to_interface.xsl | The XSL stylesheet which is used to generate the HTML file of the interface. It takes the XML source document as input as well as a few optional parameters described in HexTAC_generate_interface.php. |
| js/HexTAC_main.js | Defines global variables and initializes the Javascript environment. Some variables can be adjusted for specific needs. |
| js/HexTAC_operation.js | Deals with editing operations: drag-and-drop, deleting and sorting of sentences, undo/redo, save summary. The "Save summary" button of the interface saves the new extract in a form hidden variable and then send it back to the server. |
| js/HexTAC_sentence.js | Handles sentences as HTML elements. |
| js/jquery-1.4.2.min.js | JQuery library. |
| js/jquery-ui-1.8.custom.min.js | JQuery User Interface library, required for drag-and-drop and sorting features. |
| HexTAC.css | The CSS stylesheet of the interface. It can be edited in order to modify the page layout, including dimensions of each part of the interface, colors, sentence highlighting method, graphics of buttons. |
| COPYING | The GPL licence text. |
| example | A directory containing input and output XML example files as well as RNC files describing their format. The XML files are used by the example PHP scripts described above. More information on XML elements and attributes can be found in XML documents. |
| img | A zipped directory containing buttons' graphics. |
Author: Mehdi Yousfi-Monod (mehdi *dot* yousfi *at* gmail *dot* com)
