Post on 20-Jan-2016
EMPOWER2
Empirical Methods for Multilingual Processing,
‘Onoring Words, Enabling Rapid Ramp-up
Martha Palmer, Aravind Joshi, Mitch Marcus, Mark Liberman, Tony Kroch, Lyle Ungar
University of Pennsylvania
March 23, 2000
TIDES KICKOFF
Penn approach
• Relies on lexically based linguistic analysis– Humans annotate naturally occurring text
• (hand correct output of automatic parsers, e.g. Fiddich, XTAG)
– Train statistical POStaggers, parsers, etc.– Common thread is predicate-argument structure
Hypothesis: More linguistically sophisticated analyzers
More accurate output
EMPOWER2
Approach
• Annotations enriched with semantics and pragmatics
• Provide companion lexicons for annotated corpora• Extend our coverage to other languages
Goal – Parallel annotated corpora/lexicons will enable rapid ramp-up of MT
TOOLS/RESOURCES
• Morphological Analyzers
• Stochastic parsers
• Lexicalized grammars
• Lexical classifications for cross-lingual mappings
LANGUAGES
• English
• Chinese
• Korean
• Hindi/Tamil
Faster development of Tools/annotation:
Current Status
• English Q&A using coreference• English annotation
– adding semantics to Penn TreeBank– creating companion lexicon
• Korean/English annotation – syntactic annotation and some semantics– companion transfer lexicon
• Chinese annotation – syntactic annotation (Chinese TreeBank)
English Q&A – Tom MortonTREC-8 Approach
• Extract sentences based on:– Words in the sentence
– Category of the answer
– Words in co-reference relationships• pronouns
• common nouns
• dates
• Results – Placed 4th out of 20 participants.
Examples (pronouns)
• Who killed Lee Harvey Oswald? - demo– ..., and the hat. There was the suit he wore on the day he(JACK
RUBY) killed Oswald, a diamond-studded watch, a silver and diamond ring, two pairs of swim trunks, a shower cap, an athletic
supporter and a letter written to a woman. Other than th…
• Future Plans– use WordNet and syntactic constructions to
determine semantic categories of noun phrases– Cross-document co-reference
Semantic Annotation –Hoa Dang, Joseph Rosenzweig, John Duda
• Current syntactic annotation – POS, phrase structure bracketing– Logical Subject, locative, temporal adjuncts
• New semantic augmentations– Sense tag verbs and noun arguments/adjuncts– Predicate-argument relations for verbs, label
arguments (arg0, arg1, arg2)
First Experiment (Siglex99)• WSJ 5K word corpus
– running text– WordNet 1.6
• 2100 words sense tagged twice (10 days)– 89% inter-annotator agreement – 700 verb tokens – 81% agreement
(disagreement in 90/350 verb tokens)
• Automatic predicate-argument labeling – 81% precision on 162 structures– Hand corrected 2100 words in one day
Example
• I was shaking the whole time.
<arg0> <WN2> <temporal>
• The walls shook; the building rocked.
<arg1> <WN3>; <arg1> <WN1>
Second Experiment: Methodology(150K target – Penn TreeBank II, with Christiane Fellbaum)• Sense tagging
– Two human annotators (replace one with automatic WSD if possible)– WordNet senses, but allow for revision of entries
• Predicate argument labels – Rosenzweig’s converter– Uses TreeBank “cues”– Consults lexical semantic KB
• Verb subcategorization frames and alternations• Ontology of noun-phrase referents• Multi-word lexical items
• XML annotation in external file referencing IDs
Predicate-Argument Labeling:one raid tree – Rosenzweig’s converter
Predicate-Argument Labeling:one raid tree – Rosenzweig’s converter
New language/English MT Components
• New language– Morphological
Analyzer (POStags)
– Parser/Generator
– TreeBank
– Companion pred-arg lexicon
• English– POStagger
– Parser/Generator
– TreeBank
– Companion pred-arg lexicon
Transfer Lexicon
Korean/English MT Chunghye Han, Juntae Yoon, Meesook Kim, Eonsuk Ko
(CoGenTex/Penn/Systran: ARL)• Parallel TreeBanks for Korean/English enable
– Training of domain-specific Korean parsers• Collins parser and SuperTagger (also English)
– Alignment of Korean/English structures• Attempt automatic and semi-automatic testing and generation of transfer
lexicon (with CoGenTex)• Apply statistical MT techniques
• Lexical semantics (Systran, mapped to EuroWordNet-IL) should improve
– Accuracy of parsers – Recovery of dropped arguments
• http://www.cis.upenn.edu/~xtag/koreantag/index.html
AdditionalKorean/English parallel data?
• Current parallel corpus not public domain
• Can use tools trained on this corpus to quickly annotate additional corpora– Translate sections of Penn TreeBank into
Korean?– Use existing Korean newswire text – translate
into English?– Both?
Example translation
Transfer lexicon entries: Mapping predicate argument structures across languages
Chinese TreeBank – DODFei Xia, Ninwen Xue, Fu-dong Chiou
http://www.ldc.upenn.edu/ctb/index.html
• Workshop of interested members of Chinese community, June ‘98
• Guidelines and sample files posted on web– Segmentation, March, ‘99– POStagging, March, ‘99– Bracketing, First pass, October, ’99– Bracketing, Second Pass, May, ’00
• 95%+ inter-annotator consistency
• Release of 100K annotated data, July, ’00• Follow-up workshop, Hong Kong, ACL’00
Goal for Chinese
• Parallel, annotated corpora – Hong Kong news?
• Parse English with WSJ trained parsers, correct• Extend English TreeBank lexicon as needed• Parse Chinese with CTB trained parsers, correct• Start with lexicon extracted from CTB, extend
Experiment with using semi-automated techniques wherever possible to speed up process
Past results
• XTAG project http://www.cis.upenn.edu/~xtag/
• Penn TreeBank http://www.cis.upenn.edu/~treebank/
• Enabled the development of tools: POStaggers, parsers, co-reference, etc http://www.ircs.upenn.edu/knowledge/licensing.html