Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
Maxwell, K. Tamsin and Oberlander, Jon and Croft, W. Bruce

Article Structure

Abstract

Techniques that compare short text segments using dependency paths (or simply, paths) appear in a wide range of automated language processing applications including question answering (QA).

Introduction

In the past decade, an increasing number of techniques have used complex and effective syntactic and semantic features to determine the similarity, entailment or alignment between short texts.

Related work

Techniques that compare short text segments using dependency paths are applied to a wide range of automated language processing tasks, including paraphrasing, summarization, entailment detection, QA, machine translation and the evaluation of word, phrase and sentence similarity.

Catenae as semantic units

Catenae (Latin for ‘chain’, singular catena) are dependency-based syntactic units.

Selection method for catenae

Catenae describe relatively few of the possible word combinations in a sentence, but still include many combinations that do not result in successful ellipsis and are not informative for IR.

Experimental setup

5.1 Classification

Evaluation framework

6.1 Baseline IR models

Conclusion

We presented a flexible implementation of dependency paths for long queries in ad hoc IR that does not require dependency parsing a collection.

Topics

dependency paths

Appears in 25 sentences as: Dependency path (3) dependency path (5) Dependency paths (3) dependency paths (16)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. Techniques that compare short text segments using dependency paths (or simply, paths) appear in a wide range of automated language processing applications including question answering (QA).
    Page 1, “Abstract”
  2. In this paper, we introduce a flexible notion of paths that describe chains of words on a dependency path .
    Page 1, “Abstract”
  3. Dependency paths (or simply, paths) are compared using techniques such as tree edit distance (Punyakanok et al., 2004; Heilman and Smith, 2010), relation probability (Gao et al., 2004) and parse tree alignment (Wang et al., 2007; Park et al., 2011).
    Page 1, “Introduction”
  4. Much work on sentence similarity using dependency paths focuses on question answering (QA) where textual inference requires attention to linguistic detail.
    Page 1, “Introduction”
  5. In this paper, we explore a flexible application of dependency paths that overcomes this difficulty.
    Page 1, “Introduction”
  6. Previous heuristic filters for dependency paths (Lin and Pantel, 2001; Shen et al., 2005; Cui et al., 2005) can exclude informative relations.
    Page 1, “Introduction”
  7. These shortcomings suggest that an optimized selection process may improve performance of techniques that use dependency paths in ad hoc IR.
    Page 2, “Introduction”
  8. Techniques that compare short text segments using dependency paths are applied to a wide range of automated language processing tasks, including paraphrasing, summarization, entailment detection, QA, machine translation and the evaluation of word, phrase and sentence similarity.
    Page 2, “Related work”
  9. A generic approach uses a matching function to compare a dependency path between any two stemmed terms cc and y in a sentence A with any dependency path between cc and y in sentence B.
    Page 2, “Related work”
  10. The match score for A and B is computed over all dependency paths in A.
    Page 2, “Related work”
  11. Syntactic language models for IR are a significant departure from this trend (Gao et al., 2004; Lee et al., 2006; Cai et al., 2007; Maisonnasse et al., 2007) that use dependency paths to address long-distance dependencies and normalize spurious differences in surface text.
    Page 2, “Related work”

See all papers in Proc. ACL 2013 that mention dependency paths.

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

Back to top.

co-occurrence

Appears in 8 sentences as: Co-occurrence (3) co-occurrence (5)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. In ad hoc IR, most models of term dependence use word co-occurrence and proximity (Song and Croft, 1999; Metzler and Croft, 2005; Srikanth and Srihari, 2002; van Rijsbergen, 1993).
    Page 2, “Related work”
  2. Co-occurrence features: A governor wl tends to subcategorize for its dependents wn.
    Page 5, “Selection method for catenae”
  3. We conclude that co-occurrence is an important feature of dependency relations (Mel’cuk, 2003).
    Page 5, “Selection method for catenae”
  4. In addition, term frequencies and inverse document frequencies calculated using word co-occurrence measures are commonly used in IR.
    Page 5, “Selection method for catenae”
  5. to normalize co-occurrence counts for catenae of different lengths: a factor |c|lcl, where |c| is the number of words in catena c (Hagen et al., 2011), and the average score for a feature type over all pairwise word combinations in c.
    Page 5, “Selection method for catenae”
  6. In general, path and co-occurrence features are similar to those applied by Surdeanu et al.
    Page 5, “Selection method for catenae”
  7. Co-occurrence features (C)
    Page 6, “Experimental setup”
  8. Co-occurrence and IR effectiveness prediction features (CI) was the most influential class, and accounted for 70% of all features in the model.
    Page 7, “Experimental setup”

See all papers in Proc. ACL 2013 that mention co-occurrence.

See all papers in Proc. ACL that mention co-occurrence.

Back to top.

dependency parse

Appears in 8 sentences as: dependency parse (5) dependency parser (2) dependency parsing (1)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. These approaches are motivated by the idea that sentence meaning can be flexibly captured by the syntactic and semantic relations between words, and encoded in dependency parse tree fragments.
    Page 1, “Introduction”
  2. 1 shows a dependency parse that generates 21 catenae in total: (using 2' for Xi) 1, 2, 3, 4, 5, 6, 12, 23, 34, 45, 56, 123, 234, 345, 456, 1234, 2345, 3456, 12345, 23456, 123456.
    Page 3, “Catenae as semantic units”
  3. This highlights the fact that a single dependency parse may only partially represent the ambiguous semantics of a query.
    Page 4, “Catenae as semantic units”
  4. We use a pseudo-projective joint dependency parse and semantic role labelling system (J ohansson and
    Page 4, “Selection method for catenae”
  5. Nugues, 2008) to generate the dependency parse .
    Page 5, “Selection method for catenae”
  6. However, any dependency parser may be applied instead.
    Page 5, “Selection method for catenae”
  7. For comparison, catenae extracted from 500 queries using the Stanford dependency parser (de Mameffe et a1., 2006) overlap with 77% of catenae extracted from the same queries using the applied parser.
    Page 5, “Selection method for catenae”
  8. We presented a flexible implementation of dependency paths for long queries in ad hoc IR that does not require dependency parsing a collection.
    Page 9, “Conclusion”

See all papers in Proc. ACL 2013 that mention dependency parse.

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

Back to top.

bigrams

Appears in 5 sentences as: bigrams (5)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. Integration of the identified catenae in queries also improves IR effectiveness compared to a highly effective baseline that uses sequential bigrams with no linguistic knowledge.
    Page 2, “Introduction”
  2. The second clique contains query bigrams that match
    Page 7, “Evaluation framework”
  3. document bigrams in 2-word ordered windows (‘#I’), A2 = 0.1.
    Page 7, “Evaluation framework”
  4. The third clique uses the same bigrams as clique 2 with an 8-word unordered window (‘#uw8’), A3 = 0.05.
    Page 7, “Evaluation framework”
  5. Our reformulated model uses the same query format as SD, but the second and third cliques contain filtered catenae instead of query bigrams .
    Page 7, “Evaluation framework”

See all papers in Proc. ACL 2013 that mention bigrams.

See all papers in Proc. ACL that mention bigrams.

Back to top.

machine learning

Appears in 3 sentences as: machine learning (3)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. Informative catenae are selected using supervised machine learning with linguistically informed features and compared to both nonlinguistic terms and catenae selected heuristically with filters derived from work on paths.
    Page 1, “Abstract”
  2. We also extend previous work with development of a linguistically informed, supervised machine learning technique for selection of informative catenae.
    Page 1, “Introduction”
  3. We also develop a linguistically informed machine learning technique for catenae selection that captures both key aspects of heuristic filters, and novel characteristics of catenae and paths.
    Page 2, “Introduction”

See all papers in Proc. ACL 2013 that mention machine learning.

See all papers in Proc. ACL that mention machine learning.

Back to top.

semantic relations

Appears in 3 sentences as: semantic relations (1) semantic relationship (1) semantically related (1)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. These approaches are motivated by the idea that sentence meaning can be flexibly captured by the syntactic and semantic relations between words, and encoded in dependency parse tree fragments.
    Page 1, “Introduction”
  2. and ‘level play’ do not have an important semantic relationship relative to the query, yet these catenae are described by parent-child relations that are commonly used to filter paths in text processing applications.
    Page 2, “Introduction”
  3. This is based on the observation that semantically related words have a variety of direct and indirect relations.
    Page 3, “Related work”

See all papers in Proc. ACL 2013 that mention semantic relations.

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

Back to top.

semantic roles

Appears in 3 sentences as: semantic role (1) semantic roles (2)
In Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
  1. We use a pseudo-projective joint dependency parse and semantic role labelling system (J ohansson and
    Page 4, “Selection method for catenae”
  2. Dependency path features: Part-of-speech tags and semantic roles have been used to filter dependency paths.
    Page 5, “Selection method for catenae”
  3. This is based on work that uses paths between head nouns and verbs (Shen et al., 2005), semantic roles (Moschitti, 2008), and all dependency paths except those that occur between words in the same base chunk (e.g.
    Page 7, “Evaluation framework”

See all papers in Proc. ACL 2013 that mention semantic roles.

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

Back to top.

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