Unsupervised Semantic Role Induction via Split-Merge Clustering
Lang, Joel and Lapata, Mirella

Article Structure

Abstract

In this paper we describe an unsupervised method for semantic role induction which holds promise for relieving the data acquisition bottleneck associated with supervised role labelers.

Introduction

Recent years have seen increased interest in the shallow semantic analysis of natural language text.

Related Work

As mentioned earlier, much previous work has focused on building supervised SRL systems (Marquez et al., 2008).

Learning Setting

We follow the general architecture of supervised semantic role labeling systems.

Argument Identification

In the supervised setting, a classifier is employed in order to decide for each node in the parse tree whether it represents a semantic argument or not.

Split-Merge Role Induction

We treat role induction as a clustering problem with the goal of assigning argument instances (i.e., specific arguments occurring in an input sentence) to clusters such that these represent semantic roles.

Experimental Setup

In this section we describe how we assessed the performance of our system.

Results

Our results are summarized in Table 2.

Conclusions

In this paper we presented a novel approach to unsupervised role induction which we formulated as a clustering problem.

Topics

semantic roles

Appears in 36 sentences as: semantic role (18) Semantic roles (1) semantic roles (19)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. In this paper we describe an unsupervised method for semantic role induction which holds promise for relieving the data acquisition bottleneck associated with supervised role labelers.
    Page 1, “Abstract”
  2. We present an algorithm that iteratively splits and merges clusters representing semantic roles , thereby leading from an initial clustering to a final clustering of better quality.
    Page 1, “Abstract”
  3. By combining role induction with a rule-based component for argument identification we obtain an unsupervised end-to-end semantic role labeling system.
    Page 1, “Abstract”
  4. The term is most commonly used to describe the automatic identification and labeling of the semantic roles conveyed by sentential constituents (Gildea and J urafsky, 2002).
    Page 1, “Introduction”
  5. Semantic roles describe the relations that hold between a predicate and its arguments, abstracting over surface syntactic configurations.
    Page 1, “Introduction”
  6. window occupies different syntactic positions — it is the object of broke in sentences (la,b), and the subject in (IC) — while bearing the same semantic role , i.e., the physical object affected by the breaking event.
    Page 1, “Introduction”
  7. The semantic roles in the examples are labeled in the style of PropBank (Palmer et al., 2005), a broad-coverage human-annotated corpus of semantic roles and their syntactic realizations.
    Page 1, “Introduction”
  8. Indeed, the analysis produced by existing semantic role labelers has been shown to benefit a wide spectrum of applications ranging from information extraction (Surdeanu et al., 2003) and question answering (Shen and Lapata, 2007), to machine translation (Wu and Fung, 2009) and summarization (Melli et al., 2005).
    Page 1, “Introduction”
  9. Current approaches have high performance —a system will recall around 81% of the arguments correctly and 95% of those will be assigned a correct semantic role (see Marquez et al.
    Page 2, “Introduction”
  10. Unfortunately, the reliance on role-annotated data which is expensive and time-consuming to produce for every language and domain, presents a major bottleneck to the widespread application of semantic role labeling.
    Page 2, “Introduction”
  11. In this paper we present a simple approach to unsupervised semantic role labeling.
    Page 2, “Introduction”

See all papers in Proc. ACL 2011 that mention semantic roles.

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

Back to top.

role labeling

Appears in 11 sentences as: role labelers (2) role labeling (7) role labels (2)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. In this paper we describe an unsupervised method for semantic role induction which holds promise for relieving the data acquisition bottleneck associated with supervised role labelers .
    Page 1, “Abstract”
  2. By combining role induction with a rule-based component for argument identification we obtain an unsupervised end-to-end semantic role labeling system.
    Page 1, “Abstract”
  3. Indeed, the analysis produced by existing semantic role labelers has been shown to benefit a wide spectrum of applications ranging from information extraction (Surdeanu et al., 2003) and question answering (Shen and Lapata, 2007), to machine translation (Wu and Fung, 2009) and summarization (Melli et al., 2005).
    Page 1, “Introduction”
  4. mantic role labeling as a supervised learning problem.
    Page 2, “Introduction”
  5. Unfortunately, the reliance on role-annotated data which is expensive and time-consuming to produce for every language and domain, presents a major bottleneck to the widespread application of semantic role labeling .
    Page 2, “Introduction”
  6. In this paper we present a simple approach to unsupervised semantic role labeling .
    Page 2, “Introduction”
  7. Swier and Stevenson (2004) induce role labels with a bootstrapping scheme where the set of labeled instances is iteratively expanded using a classifier trained on previously labeled instances.
    Page 2, “Related Work”
  8. We follow the general architecture of supervised semantic role labeling systems.
    Page 3, “Learning Setting”
  9. Although the dataset provides annotations for verbal and nominal predicate-argument constructions, we only considered the former, following previous work on semantic role labeling (Marquez et al., 2008).
    Page 7, “Experimental Setup”
  10. This baseline has been previously used as point of comparison by other unsupervised semantic role labeling systems (Grenager and Manning, 2006; Lang and Lapata, 2010) and shown difficult to outperform.
    Page 7, “Experimental Setup”
  11. Coupled with a rule-based component for automatically identifying argument candidates our split-merge algorithm forms an end-to-end system that is capable of inducing role labels without any supervision.
    Page 9, “Conclusions”

See all papers in Proc. ACL 2011 that mention role labeling.

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

Back to top.

iteratively

Appears in 7 sentences as: iteratively (6) itive (1)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. We present an algorithm that iteratively splits and merges clusters representing semantic roles, thereby leading from an initial clustering to a final clustering of better quality.
    Page 1, “Abstract”
  2. Swier and Stevenson (2004) induce role labels with a bootstrapping scheme where the set of labeled instances is iteratively expanded using a classifier trained on previously labeled instances.
    Page 2, “Related Work”
  3. We formulate the induction of semantic roles as a clustering problem and propose a split-merge algorithm which iteratively manipulates clusters representing semantic roles.
    Page 3, “Related Work”
  4. Our algorithm works by iteratively splitting and merging clusters of argument instances in order to arrive at increasingly accurate representations of semantic roles.
    Page 4, “Split-Merge Role Induction”
  5. Then [3 is iteratively decreased again until it becomes zero, after which 7 is decreased by another 0.05.
    Page 6, “Split-Merge Role Induction”
  6. We proposed a split-merge algorithm that iteratively manipulates clusters representing semantic roles whilst trading off cluster purity with collocation.
    Page 9, “Conclusions”
  7. itive and requires no manual effort for training.
    Page 9, “Conclusions”

See all papers in Proc. ACL 2011 that mention iteratively.

See all papers in Proc. ACL that mention iteratively.

Back to top.

semantic role labeling

Appears in 7 sentences as: semantic role labelers (1) semantic role labeling (6)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. By combining role induction with a rule-based component for argument identification we obtain an unsupervised end-to-end semantic role labeling system.
    Page 1, “Abstract”
  2. Indeed, the analysis produced by existing semantic role labelers has been shown to benefit a wide spectrum of applications ranging from information extraction (Surdeanu et al., 2003) and question answering (Shen and Lapata, 2007), to machine translation (Wu and Fung, 2009) and summarization (Melli et al., 2005).
    Page 1, “Introduction”
  3. Unfortunately, the reliance on role-annotated data which is expensive and time-consuming to produce for every language and domain, presents a major bottleneck to the widespread application of semantic role labeling .
    Page 2, “Introduction”
  4. In this paper we present a simple approach to unsupervised semantic role labeling .
    Page 2, “Introduction”
  5. We follow the general architecture of supervised semantic role labeling systems.
    Page 3, “Learning Setting”
  6. Although the dataset provides annotations for verbal and nominal predicate-argument constructions, we only considered the former, following previous work on semantic role labeling (Marquez et al., 2008).
    Page 7, “Experimental Setup”
  7. This baseline has been previously used as point of comparison by other unsupervised semantic role labeling systems (Grenager and Manning, 2006; Lang and Lapata, 2010) and shown difficult to outperform.
    Page 7, “Experimental Setup”

See all papers in Proc. ACL 2011 that mention semantic role labeling.

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

Back to top.

CoNLL

Appears in 6 sentences as: CoNLL (6)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. Evaluation on the CoNLL 2008 benchmark dataset demonstrates that our method outperforms competitive unsupervised approaches by a wide margin.
    Page 1, “Abstract”
  2. We test the effectiveness of our induction method on the CoNLL 2008 benchmark
    Page 2, “Introduction”
  3. with the CoNLL 2008 benchmark dataset used for evaluation in our experiments.
    Page 3, “Learning Setting”
  4. Data For evaluation purposes, the system’s output was compared against the CoNLL 2008 shared task dataset (Surdeanu et al., 2008) which provides
    Page 6, “Experimental Setup”
  5. Our implementation allocates up to N = 21 clusters2 for each verb, one for each of the 20 most frequent functions in the CoNLL dataset and a default cluster for all other functions.
    Page 7, “Experimental Setup”
  6. (The following numbers are derived from the CoNLL dataset4 in the auto/auto setting.)
    Page 9, “Results”

See all papers in Proc. ACL 2011 that mention CoNLL.

See all papers in Proc. ACL that mention CoNLL.

Back to top.

gold standard

Appears in 6 sentences as: gold standard (7)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. PropBank—style gold standard annotations.
    Page 7, “Experimental Setup”
  2. In addition to gold standard dependency parses, the dataset also contains automatic parses obtained from the MaltParser (Nivre et al., 2007).
    Page 7, “Experimental Setup”
  3. Evaluation Metrics For each verb, we determine the extent to which argument instances in a cluster share the same gold standard role (purity) and the extent to which a particular gold standard role is assigned to a single cluster (collocation).
    Page 7, “Experimental Setup”
  4. 2This is the number of gold standard roles.
    Page 7, “Results”
  5. Not unexpectedly, we observe an increase in performance for all models when using gold standard parses.
    Page 8, “Results”
  6. Performance also increases if gold standard arguments are used instead of automatically identified arguments.
    Page 8, “Results”

See all papers in Proc. ACL 2011 that mention gold standard.

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

Back to top.

scoring function

Appears in 6 sentences as: Scoring Function (1) scoring function (5)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. We operationalize these notions using a scoring function that quantifies the compatibility between arbitrary cluster pairs.
    Page 3, “Related Work”
  2. Besides being inefficient, it requires a scoring function with comparable scores for arbitrary pairs of clusters.
    Page 5, “Split-Merge Role Induction”
  3. After each completion of the inner loop, the thresholds contained in the scoring function (discussed below) are adjusted and this is repeated until some termination criterion is met (discussed in Section 5.2.3).
    Page 5, “Split-Merge Role Induction”
  4. 5.2.2 Scoring Function
    Page 5, “Split-Merge Role Induction”
  5. Our scoring function quantifies whether two clusters are likely to contain arguments of the same role and was designed to reflect the following criteria:
    Page 5, “Split-Merge Role Induction”
  6. As mentioned earlier the thresholds B and y which parametrize the scoring function are adjusted at each iteration.
    Page 6, “Split-Merge Role Induction”

See all papers in Proc. ACL 2011 that mention scoring function.

See all papers in Proc. ACL that mention scoring function.

Back to top.

dependency parse

Appears in 3 sentences as: dependency parse (2) dependency parses (1)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. Given a dependency parse of a sentence, our system identifies argument instances and assigns them to clusters.
    Page 3, “Learning Setting”
  2. Figure l: A sample dependency parse with dependency labels SBJ (subject), OBJ (object), NMOD (nominal modifier), OPRD (object predicative complement), PRD (predicative complement), and IM (infinitive marker).
    Page 4, “Split-Merge Role Induction”
  3. In addition to gold standard dependency parses , the dataset also contains automatic parses obtained from the MaltParser (Nivre et al., 2007).
    Page 7, “Experimental Setup”

See all papers in Proc. ACL 2011 that mention dependency parse.

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

Back to top.

Part-of-speech

Appears in 3 sentences as: Part-of-speech (2) part-of-speech (2)
In Unsupervised Semantic Role Induction via Split-Merge Clustering
  1. When the part-of-speech similarity (pos) is below a certain threshold [3 or when clause-level constraints (cons) are satisfied to a lesser extent than threshold 7, the score takes value zero and the merge is ruled out.
    Page 6, “Split-Merge Role Induction”
  2. Part-of-speech Similarity Part-of-speech similarity is also measured through cosine-similarity (equation (3)).
    Page 6, “Split-Merge Role Induction”
  3. Clusters are again represented as vectors x and y whose components correspond to argument part-of-speech tags and values to their occurrence frequency.
    Page 6, “Split-Merge Role Induction”

See all papers in Proc. ACL 2011 that mention Part-of-speech.

See all papers in Proc. ACL that mention Part-of-speech.

Back to top.

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