Home › Forums › Calais Forums › Geek Speak (Technical usage and issues, questions, and hints ›
Named Entity Recognition Bugs
Named Entity Recognition Bugs
Posted on: Wed, 10/15/2008 - 15:49

Here is a small list of several bugs I have noticed in the Entity Recognition for Persons. I know it is a hard task, so I do not blame anyone, calais works really well. But yet, it can be improved, so here is the list :

 This is obtained by pushing the absract from the yahoo search results of "Barack Obama" to calais


    • Comments

    Another bug
    Posted on: Wed, 10/15/2008 - 15:52 #1

    I also noticed a weird thing : if i make an output text/simple "relevance" is not always an attribute
    of Persons in the xml file. I do not have the entry, sorry...
    I can provide i you ask

    Here is an example :

    Claude Zervas
    Tom Baldwin
    California
    General Hospital

    Please send us the documents
    Posted on: Wed, 10/15/2008 - 16:05 #2

    Hi,

    Please send us the documents or linked you used so we can check both issues.
    Tx,
    Ofer

    And here it is !!
    Posted on: Wed, 10/15/2008 - 16:58 #3

    In the file you can find the input and the output ( maybe not in that order, but I believe you will figure out who is who, and do not forget to watch the source ie Ctrl-U, much more readable :D ) :

    First bug : Tom Baldwin is notified as personn but without a "relevance" attribute. This sounds really like a bug so...
    http://ks36587.kimsufi.com/calaisbug/bug1.html

    Second and third bug are rather mistakes of the entity recognition tools

    "Women Voters" is not a personn ( but it is questionnable... )
    http://ks36587.kimsufi.com/calaisbug/bug2.html

    But "On Sunday" is definitely not a personn
    http://ks36587.kimsufi.com/calaisbug/bug3.html

    I found some others if you want, but I guess you'll be busy enough with these :D

    "Tom Baldwin" Issue
    Posted on: Mon, 11/17/2008 - 16:35 #4

    Hi,

    The Tom Baldwin issue is indeed a bug; thanks for noticing and letting us know! This will be fixed in the next version.

    Thanks,
    Orgad

    I forget to place the two
    Posted on: Thu, 10/16/2008 - 10:05 #5

    I forget to place the two last files. It is done.

    xml is deleted.... hope
    Posted on: Wed, 10/15/2008 - 15:53 #6

    xml is deleted.... hope you'll get it...