Skip navigation.
Home
semanticsoftware.info
Semantic Software Lab
Concordia University
Montréal, Canada
Home » Forums » Tools & Resources Forum » Semantic Software Tools and Resources

Multipax - Not Extracting many object phrases

Submitted by Anonymous on Thu, 2010-09-02 02:31
Printer-friendly versionPrinter-friendly versionPDF versionPDF version

Hi all

I used Minipar and Stanford parser. The results of multipax doesnot annotate many class. Please let me know how to fix it.

thanks
vlisr

‹ New instances populated in owlexporter NoSuchElementException when running OWLExporterDemo ›
»
rene's picture

Examples?

Submitted by rene on Fri, 2010-09-03 07:14.

I think it would help if you could provide some example sentences for each case and parser, where you are missing PASs.

»

RE: MultiPAx doesn't have much objects.

Submitted by imneal (not verified) on Sun, 2010-09-12 21:43.

Hello!

I'm also having the same problem. I'm dealing with sentences in medical texts.

for example, These sentences are not grabbed by mulitPAx at all:
"Her mother was depressed and was treated."

"FHX: Mother died of MI, age 70."

"FAMILY HISTORY: Heart disease in both grandfathers, grandmother with stroke, and a grandmother with diabetes."

All of these sentences have no object:

"She notes that a brother who was treated 12 years ago for a brain tumor has had a recurrence and had surgery again."

"Although there is no smoke exposure there is a significant family history with both Abc's father and uncle having problems with asthma as well as his older sister."

"FAMILY HX: Maternal grandfather who had a MI which she reports is secondary to tobacco and alcohol use."

"Her brother had prostate cancer."

"Brother died from pancreatitis at the age of 40 and had a prior history of mental illness."

"Brother died of Brain tumor, age 9."
"FAMILY MEDICAL HISTORY: Father had lung cancer, was smoker for 40 years."

»

object are missing

Submitted by Anonymous on Tue, 2014-10-14 01:28.

i am also facing similar problem. for example A rare black squirrel has become a regular visitor to a suburban garden
.MultiPaX 0 72 997 {dependencies=[root(17)], neg=false, obj=, parser=StanfordParserPaX, sub=squirrel, verb=visitor}

»
rene's picture

Detected by parser?

Submitted by rene on Tue, 2014-10-14 05:50.

Are you sure the object was properly detected by the parser (here Stanford)?

MultiPAX can only extract relations as they were detected by the selected parser. So you have to look at the parser's output first and check whether it created the correct subject-verb-object relations for your sentence.

Only if the relation was correctly detected by the parser, but not correctly extracted from the parser's output by MultiPAX, this is an issue we can fix.

Cheers, René

»

same here

Submitted by majermi3 on Wed, 2014-11-05 08:15.

Hi,

I am running into same problem. I use sentence from above: "A rare black squirrel has become a regular visitor to a suburban garden." and I am getting only one triple as u can see on attached picture http://postimg.org/image/n2s90ek6j/

Anybody found a solution?

Thanks.

Michal

»
rene's picture

Parser?

Submitted by rene on Wed, 2014-11-05 09:50.

Hi Michal,

Basically, I can only repeat my previous comment: Which parser did you use, and are you sure this parser actually found the relation(s) you are interested in?

You will need to look at the parser's output and check what it actually detected. If you read our MultiPaX paper: http://www.semanticsoftware.info/biblio/predicate-argument-extractor-pax you will see that different parser have very different ideas about even simple sentences!

Cheers, René

»

Hello, thanks for quick

Submitted by majermi3 on Wed, 2014-11-05 10:05.

Hello,

thanks for quick answer. I used Stanford parser and I beliave it found corect relations as it is possible to see from the picture I've added in previous post. From Stanford parser I receivedthis

  1. SyntaxTreeNode          0       21      34      {ID=34, cat=NP, consists=[30, 31, 32, 33], text=A rare black squirrel}
  2. SyntaxTreeNode          0       72      51      {ID=51, cat=ROOT, consists=[50], text=A rare black squirrel has become a regular visitor to a suburban garden.}
  3. SyntaxTreeNode          0       72      50      {ID=50, cat=S, consists=[34, 48, 49], text=A rare black squirrel has become a regular visitor to a suburban garden.}
  4. SyntaxTreeNode          0       1       30      {ID=30, cat=DT, text=A}
  5. SyntaxTreeNode          2       6       31      {ID=31, cat=JJ, text=rare}
  6. SyntaxTreeNode          7       12      32      {ID=32, cat=JJ, text=black}
  7. SyntaxTreeNode          13      21      33      {ID=33, cat=NN, text=squirrel}
  8. SyntaxTreeNode          22      25      35      {ID=35, cat=VBZ, text=has}
  9. SyntaxTreeNode          22      71      48      {ID=48, cat=VP, consists=[35, 47], text=has become a regular visitor to a suburban garden}
  10. SyntaxTreeNode          26      32      36      {ID=36, cat=VBN, text=become}
  11. SyntaxTreeNode          26      71      47      {ID=47, cat=VP, consists=[36, 40, 46], text=become a regular visitor to a suburban garden}
  12. SyntaxTreeNode          33      34      37      {ID=37, cat=DT, text=a}
  13. SyntaxTreeNode          33      50      40      {ID=40, cat=NP, consists=[37, 38, 39], text=a regular visitor}
  14. SyntaxTreeNode          35      42      38      {ID=38, cat=JJ, text=regular}
  15. SyntaxTreeNode          43      50      39      {ID=39, cat=NN, text=visitor}
  16. SyntaxTreeNode          51      53      41      {ID=41, cat=TO, text=to}
  17. SyntaxTreeNode          51      71      46      {ID=46, cat=PP, consists=[41, 45], text=to a suburban garden}
  18. SyntaxTreeNode          54      55      42      {ID=42, cat=DT, text=a}
  19. SyntaxTreeNode          54      71      45      {ID=45, cat=NP, consists=[42, 43, 44], text=a suburban garden}
  20. SyntaxTreeNode          56      64      43      {ID=43, cat=JJ, text=suburban}
  21. SyntaxTreeNode          65      71      44      {ID=44, cat=NN, text=garden}
  22. SyntaxTreeNode          71      72      49      {ID=49, cat=., text=.}

and multiPAX found this:

  1. MultiPaX        Multipax        0       72      65      {dependencies=[root(16)], neg=false, obj=, parser=StanfordParserPaX, sub=squirrel, verb=visitor}

The stenford result looks fine to me. But maybe I am wrong.

Thanks again,

Michal

»
rene's picture

GATE version?

Submitted by rene on Thu, 2014-11-06 04:10.

I've tried your example sentence, and get the PAS (squirrel, visitor, garden), which is pretty much what you'd expect.

This is with GATE-8 and the parser from the Stanford_CoreNLP plugin. Which version are you using?

Cheers, René

»

Settings

Submitted by majermi3 on Thu, 2014-11-06 04:21.

Hello,

my version is also 8 and MultiPAX 1.3. I am using Parser_Stanford from plugins

My pipeline:

  1. ANNIE English Tokeniser
  2. ANNIE Sentence Splitter
  3. ANNIE POS Tagger
  4. ANNIE Gazetteer
  5. Gate Morphological analyser
  6. StanfordParser
  7. MultiPaX

MultiPAX settings:

  1. debugFlag: false
  2. inputASName:
  3. outputAsName: Multipax
  4. parserASName: stanford
  5. raspScript: file:/D:/gate-8.0-build4825-ALL/plugins/MultiPaX-1.3/myRasp.sh

Thank you

»
rene's picture

Issue in 1.3 release

Submitted by rene on Thu, 2014-11-06 06:15.

Ok, I can reproduce your problem when using the last official release version of MultiPaX (1.3). The version we have internally has some minor changes (I'll have to investigate what exactly, since the 1.3 release has been out for quite some time).

I'll try to push out a new release, but as I'm currently traveling, it will probably not happen until the end of November. If you need it really urgently before, please send a message and we'll see what we can do.

Thanks for reporting this,

Best, René

»

Reply

Submitted by majermi3 on Thu, 2014-11-06 06:57.

Hi,

thanks for help. I don't need it now so please enjoy your traveling.

Thank for help again.

Best, Michal

»