Single Document Summarization based on Nested Tree Structure
Kikuchi, Yuta and Hirao, Tsutomu and Takamura, Hiroya and Okumura, Manabu and Nagata, Masaaki

Article Structure

Abstract

Many methods of text summarization combining sentence selection and sentence compression have recently been proposed.

Introduction

Extractive summarization is one well-known approach to text summarization and extractive methods represent a document (or a set of documents) as a set of some textual units (e.g., sentences, clauses, and words) and select their subset as a summary.

Related work

Extracting a subtree from the dependency tree of words is one approach to sentence compression (Tomita et al., 2009; Qian and Liu, 2013; Morita et al., 2013; Gillick and Favre, 2009).

Generating summary from nested tree

3.1 Building Nested Tree with RST

Experiment

4.1 Experimental Settings

Conclusion

We proposed a method of summarizing a single document that included relations between sentences and relations between words.

Topics

EDUs

Appears in 16 sentences as: EDUs (18)
In Single Document Summarization based on Nested Tree Structure
  1. Elementary Discourse Units ( EDUs ) in RST are defined as the minimal building blocks of discourse.
    Page 2, “Introduction”
  2. EDUs roughly correspond to clauses.
    Page 2, “Introduction”
  3. Most methods of summarization based on RST use EDUs as extraction textual units.
    Page 2, “Introduction”
  4. We converted the rhetorical relations between EDUs to the relations between sentences to build the nested tree structure.
    Page 2, “Introduction”
  5. A document in RST is segmented into EDUs and adjacent EDUs are linked with rhetorical relations to build an RST-Discourse Tree (RST-DT) that has a hierarchical structure of the relations.
    Page 2, “Generating summary from nested tree”
  6. RST-DT is a tree whose terminal nodes correspond to EDUs and whose nonterminal nodes indicate the relations.
    Page 2, “Generating summary from nested tree”
  7. converted RST-DTs into dependency-based discourse trees (DEP-DTs) whose nodes corresponded to EDUs and whose edges corresponded to the head modifier relationships of EDUs .
    Page 2, “Generating summary from nested tree”
  8. We specifically merge EDUs that belong to the same sentence.
    Page 2, “Generating summary from nested tree”
  9. Each sentence has only one root EDU that is the parent of all the other EDUs in the sentence.
    Page 2, “Generating summary from nested tree”
  10. We also examined the ROUGE scores of two LEAD4 methods with different textual units: EDUs (LEADEDU) and sentences (LEADSNT).
    Page 4, “Experiment”
  11. Many studies that have utilized RST have simply adopted EDUs as textual units (Mann and Thompson, 1988; Daume III and Marcu, 2002; Hirao et al., 2013; Knight and Marcu, 2000).
    Page 5, “Experiment”

See all papers in Proc. ACL 2014 that mention EDUs.

See all papers in Proc. ACL that mention EDUs.

Back to top.

subtrees

Appears in 9 sentences as: subtrees (9)
In Single Document Summarization based on Nested Tree Structure
  1. Our method jointly utilizes relations between sentences and relations between words, and extracts a rooted document subtree from a document tree whose nodes are arbitrary subtrees of the sentence tree.
    Page 2, “Introduction”
  2. However, these studies have only extracted rooted subtrees from sentences.
    Page 2, “Related work”
  3. The method of Filippova and Strube (2008) allows the model to extract non-rooted subtrees in sentence compression tasks that compress a single sentence with a given compression ratio.
    Page 2, “Related work”
  4. In particular, we extract a rooted document subtree from the document tree, and sentence subtrees from sentence trees in the document tree.
    Page 3, “Generating summary from nested tree”
  5. to extract non-rooted sentence subtrees , as we previously mentioned.
    Page 3, “Generating summary from nested tree”
  6. Constraints (6)-(10) allow the model to extract subtrees that have an arbitrary root node.
    Page 3, “Generating summary from nested tree”
  7. Rooted sentence subtree only selects rooted sentence subtrees 2.
    Page 4, “Experiment”
  8. As we can see, subtree selection only selected important subtrees that did not include the parser’s root, e.g., purpose-clauses and that-clauses.
    Page 4, “Experiment”
  9. We also discussed the effectiveness of sentence subtree selection that did not restrict rooted subtrees .
    Page 5, “Conclusion”

See all papers in Proc. ACL 2014 that mention subtrees.

See all papers in Proc. ACL that mention subtrees.

Back to top.

dependency trees

Appears in 6 sentences as: dependency tree (1) dependency tree: (1) dependency trees (4)
In Single Document Summarization based on Nested Tree Structure
  1. recently transformed RST trees into dependency trees and used them for single document summarization (Hirao et al., 2013).
    Page 1, “Introduction”
  2. They formulated the summarization problem as a tree knapsack problem with constraints represented by the dependency trees .
    Page 1, “Introduction”
  3. Extracting a subtree from the dependency tree of words is one approach to sentence compression (Tomita et al., 2009; Qian and Liu, 2013; Morita et al., 2013; Gillick and Favre, 2009).
    Page 2, “Related work”
  4. Fortunately we can simply convert DEP-DTs to obtain dependency trees between sentences.
    Page 2, “Generating summary from nested tree”
  5. After the document tree is obtained, we use a dependency parser to obtain the syntactic dependency trees of sentences.
    Page 3, “Generating summary from nested tree”
  6. We added two types of constraints to our model to extract a grammatical sentence subtree from a dependency tree:
    Page 3, “Generating summary from nested tree”

See all papers in Proc. ACL 2014 that mention dependency trees.

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

Back to top.

sentence compression

Appears in 6 sentences as: sentence compression (6)
In Single Document Summarization based on Nested Tree Structure
  1. Many methods of text summarization combining sentence selection and sentence compression have recently been proposed.
    Page 1, “Abstract”
  2. There has recently been increasing attention focused on approaches that jointly optimize sentence extraction and sentence compression (Tomita et al., 2009;
    Page 1, “Introduction”
  3. Extracting a subtree from the dependency tree of words is one approach to sentence compression (Tomita et al., 2009; Qian and Liu, 2013; Morita et al., 2013; Gillick and Favre, 2009).
    Page 2, “Related work”
  4. The method of Filippova and Strube (2008) allows the model to extract non-rooted subtrees in sentence compression tasks that compress a single sentence with a given compression ratio.
    Page 2, “Related work”
  5. However, introducing sentence compression to the system greatly improved the ROUGE score (0.354).
    Page 4, “Experiment”
  6. Hence, utilizing these for sentence compression has been left for future work.
    Page 5, “Conclusion”

See all papers in Proc. ACL 2014 that mention sentence compression.

See all papers in Proc. ACL that mention sentence compression.

Back to top.

dependency parser

Appears in 3 sentences as: dependency parser (3)
In Single Document Summarization based on Nested Tree Structure
  1. We propose a method of summarizing a single document that utilizes dependency between sentences obtained from rhetorical structures and dependency between words obtained from a dependency parser .
    Page 1, “Introduction”
  2. The sentence tree is a tree that has words as nodes and head modifier relationships between words obtained by the dependency parser as edges.
    Page 1, “Introduction”
  3. After the document tree is obtained, we use a dependency parser to obtain the syntactic dependency trees of sentences.
    Page 3, “Generating summary from nested tree”

See all papers in Proc. ACL 2014 that mention dependency parser.

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

Back to top.

discourse structure

Appears in 3 sentences as: discourse structure (3)
In Single Document Summarization based on Nested Tree Structure
  1. It is important for generated summaries to have a discourse structure that is similar to that of the source document.
    Page 1, “Introduction”
  2. Rhetorical Structure Theory (RST) (Mann and Thompson, 1988) is one way of introducing the discourse structure of a document to a summarization task (Marcu, 1998; Daume III and Marcu, 2002; Hirao et al., 2013).
    Page 1, “Introduction”
  3. The nucleus is more salient to the discourse structure , while the other span, the satellite, represents supporting information.
    Page 2, “Generating summary from nested tree”

See all papers in Proc. ACL 2014 that mention discourse structure.

See all papers in Proc. ACL that mention discourse structure.

Back to top.

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