<div dir="ltr"><div>Hi Fatemeh,</div><div><br></div>On Wed, Dec 18, 2013 at 4:47 AM, Fatemeh Torabi Asr <span dir="ltr"><<a href="mailto:torabiasr@gmail.com" target="_blank">torabiasr@gmail.com</a>></span> wrote:<br>
<div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr">
<div>Hi all,<br><br></div>Does anybody know of the latest [state-of-the-art] [efficient] [open source] package or toolkit for detecting events plus whatever within document information about them (the more the better). <br>

<br>Most favorable example which I copy from FASTUS old paper:<br><br><font size="3"><pre><b>Incident: Date</b>                   - 19 Apr 89
<b>Incident: Location</b>               El Salvador: San Salvador (CITY)
<b>Incident: Type</b>                   Bombing
<b>Perpetrator: Individual ID</b>      "urban guerrillas"
<b>Perpetrator: Organization ID</b>     "FMLN"
<b>Perpetrator: Organization</b>        Suspected or Accused by Authorities:  "FMLN"
               <b>Confidence</b>        
<b>Physical Target: Description</b>     "vehicle"
<b>Physical Target: Effect</b>          Some Damage:  "vehicle"
<b>Human Target: Name</b>               "Roberto Garcia Alvarado"
<b>Human Target: Description</b>        "attorney general": "Roberto Garcia Alvarado"
                                 "driver"
                                 "bodyguards"
<b>Human Target: Effect</b>             Death: "Roberto Garcia Alvarado"
                                 No Injury: "driver"
                                 Injury: "bodyguards"</pre></font></div></blockquote><div>This sort of extraction is very task-dependent (do you actually want to extract Latin American terrorist incidents?), and I'm not sure what's available as software.</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr"><font size="3"><pre><font size="3"><font size="3"><font size="3">A</font> second related question:</font> <font size="3">i</font></font>s T<font size="3">ARSQI the be<font size="3">st<font size="3"> implementation of the T<font size="3">imeML which<font size="3"> is readily available</font>?</font></font></font></font></pre>
</font></div></blockquote><div>No. Some of the systems that performed well at the recent TempEval evaluations of TimeML technology are available online:</div><div><ul><li>TIMEE/TIPSem by Hector Llorens: <a href="http://gplsi.dlsi.ua.es/demos/TIMEE/">http://gplsi.dlsi.ua.es/demos/TIMEE/</a> (binary release)<br>
</li><li>ClearTk,including Steven Bethard's TimeML tagger: <a href="https://code.google.com/p/cleartk/">https://code.google.com/p/cleartk/</a> (open-source, built on UIMA)</li></ul><div>Other tools are available for individual components of the task, such as temporal expression recognition and normalisation (e.g. <a href="https://code.google.com/p/heideltime/">https://code.google.com/p/heideltime/</a>).</div>
</div><div><br></div><div>Cheers,</div><div><br></div><div>Joel Nothman</div></div></div><div class="gmail_extra">Schwa Lab</div><div class="gmail_extra">School of IT</div><div class="gmail_extra">University of Sydney</div>
</div>