Fully Unsupervised Core-Adjunct Argument Classification
Abend, Omri and Rappoport, Ari

Article Structure

Abstract

The core-adjunct argument distinction is a basic one in the theory of argument structure.

Introduction

The distinction between core arguments (henceforth, cores) and adjuncts is included in most theories on argument structure (Dowty, 2000).

Core-Adjunct in Previous Work

PropBank.

Algorithm

We are given a (predicate, argument) pair in a test sentence, and we need to determine whether the argument is a core or an adjunct.

Experimental Setup

Experiments were conducted in two scenarios.

Results

Table 1 presents the results of our main experiments.

Conclusion

We presented a fully unsupervised algorithm for the classification of arguments into cores and adjuncts.

Topics

POS tagger

Appears in 11 sentences as: POS tag (1) POS tagger (5) POS taggers (2) POS tagging (3) POS tags (1)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. However, no work has tack-lwnyamammemmfimdmmmm.Umw pervised models reduce reliance on the costly and error prone manual multilayer annotation ( POS tagging , parsing, core-adjunct tagging) commonly used for this task.
    Page 1, “Introduction”
  2. In addition, supervised models utilize supervised parsers and POS taggers, while the current state-of-the-art in unsupervised parsing and POS tagging is considerably worse than their supervised counterparts.
    Page 2, “Core-Adjunct in Previous Work”
  3. First, all works use manual or supervised syntactic annotations, usually including a POS tagger .
    Page 3, “Core-Adjunct in Previous Work”
  4. To estimate this joint distribution, PSH samples are extracted from the training corpus using unsupervised POS taggers (Clark, 2003; Abend et al., 2010) and an unsupervised parser (Seginer, 2007).
    Page 4, “Algorithm”
  5. This parser is unique in its ability to induce a bracketing (unlabeled parsing) from raw text (without even using POS tags ) with strong results.
    Page 4, “Algorithm”
  6. We continue by tagging the corpus using Clark’s unsupervised POS tagger (Clark, 2003) and the unsupervised Prototype Tagger (Abend et al., 2010)2.
    Page 4, “Algorithm”
  7. The corpus is now tagged using an unsupervised POS tagger .
    Page 4, “Algorithm”
  8. Each word in the argument is now represented by its word form (without lemmatization), its unsupervised POS tag and its depth in the parse tree of the argument.
    Page 5, “Algorithm”
  9. This scenario decouples the accuracy of the algorithm from the quality of the unsupervised POS tagging .
    Page 6, “Experimental Setup”
  10. Finally, we experiment on a scenario where even argument identification on the test set is not provided, but performed by the algorithm of (Abend et al., 2009), which uses neither syntactic nor SRL annotation but does utilize a supervised POS tagger .
    Page 8, “Experimental Setup”
  11. The algorithm applies state-of-the-art unsupervised parser and POS tagger to collect statistics from a large raw text corpus.
    Page 9, “Conclusion”

See all papers in Proc. ACL 2010 that mention POS tagger.

See all papers in Proc. ACL that mention POS tagger.

Back to top.

semantic role

Appears in 7 sentences as: Semantic Role (1) semantic role (5) semantic roles (1)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. The task of distinguishing between the two has strong relations to various basic NLP tasks such as syntactic parsing, semantic role labeling and subcategorization acquisition.
    Page 1, “Abstract”
  2. Adjuncts form an independent semantic unit and their semantic role can often be inferred independently of the predicate (e.g., [after lunch] is usually a temporal modifier).
    Page 1, “Introduction”
  3. Distinguishing between the two argument types has been discussed extensively in various formulations in the NLP literature, notably in PP attachment, semantic role labeling (SRL) and subcategorization acquisition.
    Page 1, “Introduction”
  4. It takes a different approach from PB to semantic roles .
    Page 2, “Core-Adjunct in Previous Work”
  5. It does not commit that a semantic role is consistently tagged as a core or a non-core across frames.
    Page 2, “Core-Adjunct in Previous Work”
  6. For example, the semantic role ‘path’ is considered core in the ‘Self Motion’ frame, but as non-core in the ‘Placing’ frame.
    Page 2, “Core-Adjunct in Previous Work”
  7. Semantic Role Labeling.
    Page 2, “Core-Adjunct in Previous Work”

See all papers in Proc. ACL 2010 that mention semantic role.

See all papers in Proc. ACL that mention semantic role.

Back to top.

gold standard

Appears in 3 sentences as: gold standard (3)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. In the ‘SID’ (supervised identification of prepositions and verbs) scenario, a gold standard list of prepositions was provided.
    Page 6, “Experimental Setup”
  2. The list was generated by taking every word tagged by the preposition tag (‘IN’) in at least one of its instances under the gold standard annotation of the WSJ sections 2—2l.
    Page 6, “Experimental Setup”
  3. An unlabeled match is defined to be an argument that agrees in its boundaries with a gold standard argument and a labeled match requires in addition that the arguments agree in their core/adjunct label.
    Page 8, “Experimental Setup”

See all papers in Proc. ACL 2010 that mention gold standard.

See all papers in Proc. ACL that mention gold standard.

Back to top.

parse tree

Appears in 3 sentences as: parse tree (3)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. A sequence of words will be marked as an argument of the verb if it is a constituent that does not contain the verb (according to the unsupervised parse tree ), whose parent is an ancestor of the verb.
    Page 4, “Algorithm”
  2. Each word in the argument is now represented by its word form (without lemmatization), its unsupervised POS tag and its depth in the parse tree of the argument.
    Page 5, “Algorithm”
  3. Instead, only those appearing in the top level (depth = l) of the argument under its unsupervised parse tree are taken.
    Page 5, “Algorithm”

See all papers in Proc. ACL 2010 that mention parse tree.

See all papers in Proc. ACL that mention parse tree.

Back to top.

role labeling

Appears in 3 sentences as: Role Labeling (1) role labeling (2)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. The task of distinguishing between the two has strong relations to various basic NLP tasks such as syntactic parsing, semantic role labeling and subcategorization acquisition.
    Page 1, “Abstract”
  2. Distinguishing between the two argument types has been discussed extensively in various formulations in the NLP literature, notably in PP attachment, semantic role labeling (SRL) and subcategorization acquisition.
    Page 1, “Introduction”
  3. Semantic Role Labeling .
    Page 2, “Core-Adjunct in Previous Work”

See all papers in Proc. ACL 2010 that mention role labeling.

See all papers in Proc. ACL that mention role labeling.

Back to top.

semantic role labeling

Appears in 3 sentences as: Semantic Role Labeling (1) semantic role labeling (2)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. The task of distinguishing between the two has strong relations to various basic NLP tasks such as syntactic parsing, semantic role labeling and subcategorization acquisition.
    Page 1, “Abstract”
  2. Distinguishing between the two argument types has been discussed extensively in various formulations in the NLP literature, notably in PP attachment, semantic role labeling (SRL) and subcategorization acquisition.
    Page 1, “Introduction”
  3. Semantic Role Labeling .
    Page 2, “Core-Adjunct in Previous Work”

See all papers in Proc. ACL 2010 that mention semantic role labeling.

See all papers in Proc. ACL that mention semantic role labeling.

Back to top.

similarity measure

Appears in 3 sentences as: similarity measure (3)
In Fully Unsupervised Core-Adjunct Argument Classification
  1. sim(h, h’) is a similarity measure between argument heads.
    Page 5, “Algorithm”
  2. The similarity measure we use is based on the slot distributions of the arguments.
    Page 5, “Algorithm”
  3. The similarity measure between two head words is then defined as the cosine measure of their vectors.
    Page 5, “Algorithm”

See all papers in Proc. ACL 2010 that mention similarity measure.

See all papers in Proc. ACL that mention similarity measure.

Back to top.

PDF available from www.aclweb.org.
Send us an email