Unsupervised Ontology Induction from Text
Poon, Hoifung and Domingos, Pedro

Article Structure

Abstract

Extracting knowledge from unstructured text is a longstanding goal of NLP.

Introduction

Knowledge acquisition has been a major goal of NLP since its early days.

Background 2.1 Ontology Learning

In general, ontology induction (constructing an ontology) and ontology population (mapping textual expressions to concepts and relations in the ontology) remain difficult open problems (Staab and Studer, 2004).

Unsupervised Ontology Induction with Markov Logic

A major limitation of USP is that it either merges two object clusters into one, or leaves them separate.

Experiments

4.1 Methodology

Conclusion

This paper introduced OntoUSP, the first unsupervised end-to-end system for ontology induction and knowledge extraction from text.

Topics

semantic parses

Appears in 27 sentences as: semantic parse (8) semantic parser (4) semantic parses (13) Semantic Parsing (1) Semantic parsing (1) semantic parsing (5)
In Unsupervised Ontology Induction from Text
  1. OntoUSP builds on the USP unsupervised semantic parser by jointly forming ISA and IS-PART hierarchies of lambda-form clusters.
    Page 1, “Abstract”
  2. It has been successfully applied to unsupervised learning for various NLP tasks such as coreference resolution (Poon and Domingos, 2008) and semantic parsing (Poon and Domingos, 2009).
    Page 2, “Background 2.1 Ontology Learning”
  3. 2.3 Unsupervised Semantic Parsing
    Page 2, “Background 2.1 Ontology Learning”
  4. Semantic parsing aims to obtain a complete canonical meaning representation for input sentences.
    Page 2, “Background 2.1 Ontology Learning”
  5. It can be viewed as a structured prediction problem, where a semantic parse is formed by partitioning the input sentence (or a syntactic analysis such as a dependency tree) into meaning units and assigning each unit to the logical form representing an entity or relation (Figure 1).
    Page 2, “Background 2.1 Ontology Learning”
  6. Figure 1: An example of semantic parsing .
    Page 3, “Background 2.1 Ontology Learning”
  7. Top: semantic parsing converts an input sentence into logical form in Davidsonian semantics.
    Page 3, “Background 2.1 Ontology Learning”
  8. Bottom: a semantic parse consists of a partition of the dependency tree and an assignment of its parts.
    Page 3, “Background 2.1 Ontology Learning”
  9. parser extracts knowledge from input text and converts them into logical form (the semantic parse ), which can then be used in logical and probabilistic inference and support end tasks such as question answering.
    Page 3, “Background 2.1 Ontology Learning”
  10. A major challenge to semantic parsing is syntactic and lexical variations of the same meaning, which abound in natural languages.
    Page 3, “Background 2.1 Ontology Learning”
  11. ).3 For each sentence, a semantic parse comprises of a partition of its QLF into subexpressions, each of which has a naturally corresponding lambda
    Page 3, “Background 2.1 Ontology Learning”

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

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

Back to top.

dependency trees

Appears in 7 sentences as: dependency tree (3) dependency trees (4)
In Unsupervised Ontology Induction from Text
  1. It can be viewed as a structured prediction problem, where a semantic parse is formed by partitioning the input sentence (or a syntactic analysis such as a dependency tree ) into meaning units and assigning each unit to the logical form representing an entity or relation (Figure 1).
    Page 2, “Background 2.1 Ontology Learning”
  2. Bottom: a semantic parse consists of a partition of the dependency tree and an assignment of its parts.
    Page 3, “Background 2.1 Ontology Learning”
  3. Recently, we developed the USP system (Poon and Domingos, 2009), the first unsupervised approach for semantic parsing.2 USP inputs dependency trees of sentences and first transforms them into quasi-logical forms (QLFs) by converting each node to a unary atom and each dependency edge to a binary atom (e.g., the node for “induces” becomes induces(e1) and the subject dependency becomes nsubj(e1, e2), where ei’s are Skolem constants indexed by the nodes.
    Page 3, “Background 2.1 Ontology Learning”
  4. This novel form of relational clustering is governed by a joint probability distribution P (T, L) defined in higher-orde15 Markov logic, Where T are the input dependency trees , and L the semantic parses.
    Page 3, “Background 2.1 Ontology Learning”
  5. Given the dependency tree T of a sentence, the conditional probability of a semantic parse L is given by P7“(L|T) oc exp (2, wini(T, The MAP semantic parse is simply
    Page 5, “Unsupervised Ontology Induction with Markov Logic”
  6. OntoUSP uses the same learning objective as USP, i.e., to find parameters 6 that maximizes the log-likelihood of observing the dependency trees T, summing out the unobserved semantic parses L:
    Page 5, “Unsupervised Ontology Induction with Markov Logic”
  7. USP (Poon and Domingos, 2009) parses the input text using the Stanford dependency parser (Klein and Manning, 2003; de Marneffe et al., 2006), learns an MLN for semantic parsing from the dependency trees , and outputs this MLN and the MAP semantic parses of the input sentences.
    Page 8, “Experiments”

See all papers in Proc. ACL 2010 that mention dependency trees.

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

Back to top.

logical form

Appears in 5 sentences as: logical form (5)
In Unsupervised Ontology Induction from Text
  1. We propose OntoUSP (Ontological USP), a system that learns an ISA hierarchy over clusters of logical expressions, and populates it by translating sentences to logical form .
    Page 1, “Introduction”
  2. It can be viewed as a structured prediction problem, where a semantic parse is formed by partitioning the input sentence (or a syntactic analysis such as a dependency tree) into meaning units and assigning each unit to the logical form representing an entity or relation (Figure 1).
    Page 2, “Background 2.1 Ontology Learning”
  3. Top: semantic parsing converts an input sentence into logical form in Davidsonian semantics.
    Page 3, “Background 2.1 Ontology Learning”
  4. parser extracts knowledge from input text and converts them into logical form (the semantic parse), which can then be used in logical and probabilistic inference and support end tasks such as question answering.
    Page 3, “Background 2.1 Ontology Learning”
  5. 3W6 call these QLFs because they are not true logical form (the ambiguities are not yet resolved).
    Page 3, “Background 2.1 Ontology Learning”

See all papers in Proc. ACL 2010 that mention logical form.

See all papers in Proc. ACL that mention logical form.

Back to top.

end-to-end

Appears in 4 sentences as: end-to-end (4)
In Unsupervised Ontology Induction from Text
  1. Although learning approaches to many of its subtasks have been developed (e.g., parsing, taxonomy induction, information extraction), all end-to-end solutions to date require heavy supervision and/or manual engineering, limiting their scope and scalability.
    Page 1, “Abstract”
  2. (2006)), but to date there is no sufficiently automatic end-to-end solution.
    Page 1, “Introduction”
  3. Ideally, we would like to have an end-to-end unsupervised (or lightly supervised) solution to the problem of knowledge acquisition from text.
    Page 1, “Introduction”
  4. This paper introduced OntoUSP, the first unsupervised end-to-end system for ontology induction and knowledge extraction from text.
    Page 9, “Conclusion”

See all papers in Proc. ACL 2010 that mention end-to-end.

See all papers in Proc. ACL that mention end-to-end.

Back to top.

knowledge base

Appears in 4 sentences as: knowledge base (2) knowledge bases (2)
In Unsupervised Ontology Induction from Text
  1. We evaluate OntoUSP by using it to extract a knowledge base from biomedical abstracts and answer questions.
    Page 1, “Abstract”
  2. Besides, many of them either bootstrap from heuristic patterns (e.g., Hearst patterns (Hearst, 1992)) or build on existing structured or semistructured knowledge bases (e.g., WordNet (Fellbaum, 1998) and Wikipedial), thus are limited in coverage.
    Page 2, “Background 2.1 Ontology Learning”
  3. Our approach can also leverage existing ontologies and knowledge bases to conduct semi-supervised ontology induction (e.g., by incorporating existing structures as hard constraints or penalizing deviation from them).
    Page 2, “Background 2.1 Ontology Learning”
  4. These MAP parses formed the knowledge base (KB).
    Page 8, “Experiments”

See all papers in Proc. ACL 2010 that mention knowledge base.

See all papers in Proc. ACL that mention knowledge base.

Back to top.

question answering

Appears in 4 sentences as: question answering (4)
In Unsupervised Ontology Induction from Text
  1. Finally, experiments on a biomedical knowledge acquisition and question answering task show that OntoUSP can greatly outperform USP and previous systems.
    Page 2, “Introduction”
  2. parser extracts knowledge from input text and converts them into logical form (the semantic parse), which can then be used in logical and probabilistic inference and support end tasks such as question answering .
    Page 3, “Background 2.1 Ontology Learning”
  3. Table 1: Comparison of question answering results on the GENIA dataset.
    Page 8, “Experiments”
  4. To use DIRT in question answering , it was queried to obtain similar paths for the relation of the question, which were then used to match sentences.
    Page 8, “Experiments”

See all papers in Proc. ACL 2010 that mention question answering.

See all papers in Proc. ACL that mention question answering.

Back to top.

learning algorithm

Appears in 3 sentences as: learning algorithm (2) learning algorithms (1)
In Unsupervised Ontology Induction from Text
  1. We then present the OntoUSP Markov logic network and the inference and learning algorithms used with it.
    Page 2, “Introduction”
  2. Finally, we describe the learning algorithm and how OntoUSP induces the ontology while learning the semantic parser.
    Page 5, “Unsupervised Ontology Induction with Markov Logic”
  3. Algorithm 2 gives pseudo-code for OntoUSP’s learning algorithm .
    Page 6, “Unsupervised Ontology Induction with Markov Logic”

See all papers in Proc. ACL 2010 that mention learning algorithm.

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

Back to top.