---
title: "Building a Semantic Lexicon of English Nouns via Bootstrapping"
description: "The authors propose a weakly supervised bootstrapping method that grows a semantic class from a small seed list. 12 references."
authors: ["Ting Qian", "Benjamin Van Durme", "Lenhart Schubert"]
year: 2009
venue: "Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Student Research Workshop and Doctoral Consortium"
source: "https://aclanthology.org/N09-3007/"
licence: "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)"
licence_url: "https://creativecommons.org/licenses/by-nc-sa/3.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/qian-2009-building-semantic-lexicon-english-nouns/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/qian-2009-building-semantic-lexicon-english-nouns/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2009](https://md.research.elamur.ai/free/en/papers/2009/page/2/) / Building a Semantic Lexicon of English Nouns via Bootstrapping

# Building a Semantic Lexicon of English Nouns via Bootstrapping

**Authors:** Ting Qian, [Benjamin Van Durme](https://md.research.elamur.ai/free/en/authors/benjamin-van-durme/) and [Lenhart Schubert](https://md.research.elamur.ai/free/en/authors/lenhart-schubert/) · **Year:** 2009 · **Venue:** [Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Student Research Workshop and Doctoral Consortium](https://md.research.elamur.ai/free/en/venues/human-language-technologies-the-annual-conference-of-the-north-american-chapter/) ([Human Language Technologies: The Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Student Research Workshop and Doctoral Consortium 2009](https://md.research.elamur.ai/free/en/venues/human-language-technologies-the-annual-conference-of-the-north-american-chapter/2009/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY-NC-SA 3.0 (ACL Anthology, before 2016)](https://creativecommons.org/licenses/by-nc-sa/3.0/)

**Links:** [Publisher's page](https://aclanthology.org/N09-3007/) · [PDF](https://aclanthology.org/N09-3007.pdf) · [OpenAlex](https://openalex.org/W2029044131)

**Citations:** 4 citations (OpenAlex, 2026-10-14)

## Summary
The paper tackles the cost of hand-building semantic noun lexicons such as WordNet. Such lexicons are slow to extend to new words and may not fit a particular domain. The authors propose a weakly supervised bootstrapping method that grows a semantic class from a small seed list. A perceptron classifier uses two kinds of features. Morphological cues are affixes and binned word length. Contextual cues are KNEXT "factoids", general-knowledge propositions extracted from parsed British National Corpus text, with the target noun blanked out. In each round, newly found words enter the training set only if their activation beats the mean for their category. Because the expanded set has no gold standard, the authors introduce a Bootstrapping Score to pick the best iteration after the fact. On the event-noun task over WordNet's noun list, the best model reached a score of 90.8%. A second experiment, physical versus non-physical objects, gave about 4% sampled error. This suggests the method generalizes beyond KNEXT's built-in event heuristics. It offers a cheap path to adaptable, corpus-grounded noun classes.

## Key points
- Source lexicon: the WordNet noun list (21,512 nouns). The event-noun seeds were 15 event nouns and 215 non-event nouns. Most non-event seeds were words whose endings misleadingly look eventive.
- Two feature-frequency cut-offs were tried: at least 10 occurrences (55,354 features) and at least 15 (35,457), each run for 20 iterations. Trial 2 was better on average.
- The best model, iteration 6 of Trial 2, found 236 correct and 32 wrong event nouns (11.9% error, BS 90.8%). By iteration 10 the error had climbed to 26.2% and BS had dropped to 83.8%.
- Bootstrapping Score is the harmonic mean of precision and Bootstrapping Rate, where BR = |NEW| / (|NEW| + |SEED|). It stands in for recall, since the true size of the class is unknown.
- All top-ranked features of the best model were contextual. Later models that leaned on morphology did worse.
- The physical-object run started from 15 physical and 34 non-physical seeds. It levelled off near 2,200 physical nouns by iteration 9, with about 4% error on five random 20-word samples.
- The method also found event nouns missing from WordNet's event subtree (7,655 terms). Manual analysis and k-means clustering of the best model's features were used to propose event subtypes.

## Interesting ideas
- Admit self-labelled examples only when their score is above the current mean activation of their class, so the threshold moves with model quality instead of being a fixed constant.
- Choose the stopping iteration with a harmonic mean of precision (from a small manual check) and a "growth over seed" ratio, which avoids needing recall when the full class is unknown.
- Fill the negative seeds with deliberate counterexamples to a shallow cue, such as eventive-looking suffixes on non-events, so the learner cannot shortcut through surface morphology.
- Represent a noun by the general propositions it occurs in, with the noun itself blanked out (e.g. "property may undergo a ___"), so that words sharing propositional slots cluster together.
- Rerun the method on a second distinction where the feature source has no built-in bias, to check that success is not inherited from the upstream extractor.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (learn): Silver-data admission could use a per-category, per-round mean-confidence cutoff rather than a global threshold, which may limit drift when the witness parsers or teacher tiers relabel data iteratively.
- **I2** (learn): Self-training and repair-rule induction loops could log a similar score per round and roll back to the best round, instead of trusting the final one.
- **I3** (world): When inducing noun classes (event, physical object) for the ontology, seed lists should include suffix-misleading nouns so that classification rests on usage rather than spelling.
- **I4** (world): Slots from the frames and world layer (what can undergo, be owned, be young) could serve as features for labelling unknown nouns as events or objects, helping cases like "a man with hobbies" vs "with apples".
- **I5** (learn): Every classifier trained on features produced by another Mova component should get such a control task, as a form of negative control against leakage.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [University of Rochester](https://md.research.elamur.ai/free/en/universities/university-of-rochester/) — University, Rochester, United States · 43 works · 7,874 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works

🧭 Topics (1): [English](https://md.research.elamur.ai/free/en/topics/english/) · 2,379 works

## Cite

Cite as: Ting Qian, Benjamin Van Durme and Lenhart Schubert. 2009. Building a Semantic Lexicon of English Nouns via Bootstrapping. Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Student Research Workshop and Doctoral Consortium.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{qian2009building,
  title = {{Building a Semantic Lexicon of English Nouns via Bootstrapping}},
  author = {Ting Qian, Benjamin {Van Durme} and Lenhart Schubert},
  booktitle = {Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Student Research Workshop and Doctoral Consortium},
  year = {2009},
  doi = {10.3115/1620932.1620939},
  url = {https://doi.org/10.3115/1620932.1620939}
}
```

- [← previous Arabic Disambiguation Using Dependency Grammar](https://md.research.elamur.ai/free/en/papers/daoud-2009-arabic-disambiguation-dependency-grammar/)
- [next → Unsupervised Concept Discovery In Hebrew Using Simple Unsupervised Word Prefix Segmentation for Hebrew and Arabic](https://md.research.elamur.ai/free/en/papers/dinur-2009-unsupervised-concept-discovery-hebrew-simple/)
- [↑ back to papers of 2009](https://md.research.elamur.ai/free/en/papers/2009/page/2/)

## References

12 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. BNC Consortium. 2001. The British National Corpus, version 2 (BNC World). Distributed by Oxford University Computing Services. [find](https://scholar.google.com/scholar?q=BNC+Consortium%2E+2001%2E+The+British+National+Corpus%2C+version+2+%28BNC+World%29%2E+Distributed+by+Oxford+University+Computing+Services%2E)
2. Christiane Fellbaum, editor. 1998. WordNet: An Electronic Lexical Database. MIT Press. [find](https://scholar.google.com/scholar?q=Christiane+Fellbaum%2C+editor%2E+1998%2E+WordNet%3A+An+Electronic+Lexical+Database%2E+MIT+Press%2E)
3. Marti A. Hearst. 1998. Automated discovery of Word-Net relations. In (Fellbaum, 1998), pages 131–153. [find](https://scholar.google.com/scholar?q=Marti+A%2E+Hearst%2E+1998%2E+Automated+discovery+of+Word%2DNet+relations%2E+In+%28Fellbaum%2C+1998%29%2C+pages+131%E2%80%93153%2E)
4. Donald Hindle. 1990. [Noun classification from predicate-argument structures](https://md.research.elamur.ai/free/en/papers/hindle-1990-noun-classification-predicate-argument-structures/). In ACL.
5. Alexandre Klementiev and Dan Roth. 2006. Weakly supervised named entity transliteration and discovery from multilingual comparable corpora. In ACL. [find](https://scholar.google.com/scholar?q=Alexandre+Klementiev+and+Dan+Roth%2E+2006%2E+Weakly+supervised+named+entity+transliteration+and+discovery+from+multilingual+comparable+corpora%2E+In+ACL%2E)
6. Maria Liakata and Stephen Pulman. 2008. Automatic Fine-Grained Semantic Classification for Domain Adaption. In Proceedings of Semantics in Text Processing (STEP). [find](https://scholar.google.com/scholar?q=Maria+Liakata+and+Stephen+Pulman%2E+2008%2E+Automatic+Fine%2DGrained+Semantic+Classification+for+Domain+Adaption%2E+In+Proceedings+of+Semantics+in+Text+Processing+%28STEP%29%2E)
7. Dekang Lin and Patrick Pantel. 2001. Induction of semantic classes from natural language text. In KDD. [find](https://scholar.google.com/scholar?q=Dekang+Lin+and+Patrick+Pantel%2E+2001%2E+Induction+of+semantic+classes+from+natural+language+text%2E+In+KDD%2E)
8. Brian Roark and Eugene Charniak. 1998. Noun-phrase co-occurrence statistics for semi-automatic semantic lexicon construction. In ACL, pages 1110–1116. [find](https://scholar.google.com/scholar?q=Brian+Roark+and+Eugene+Charniak%2E+1998%2E+Noun%2Dphrase+co%2Doccurrence+statistics+for+semi%2Dautomatic+semantic+lexicon+construction%2E+In+ACL%2C+pages+1110%E2%80%931116%2E)
9. Frank Rosenblatt. 1958. The perceptron: A probabilistic model for information storage and organization in the brain. Psychological Review, 65(6):386–408. [find](https://scholar.google.com/scholar?q=Frank+Rosenblatt%2E+1958%2E+The+perceptron%3A+A+probabilistic+model+for+information+storage+and+organization+in+the+brain%2E+Psychological+Review%2C+65%286%29%3A386%E2%80%93408%2E)
10. Lenhart K. Schubert. 2002. Can we derive general world knowledge from text? In HLT. [find](https://scholar.google.com/scholar?q=Lenhart+K%2E+Schubert%2E+2002%2E+Can+we+derive+general+world+knowledge+from+text%3F+In+HLT%2E)
11. Michael Thelen and Ellen Riloff. 2002. [A bootstrapping method for learning semantic lexicons using extraction pattern contexts](https://md.research.elamur.ai/free/en/papers/thelen-2002-bootstrapping-method-learning-semantic-lexicons/). In EMNLP.
12. Benjamin Van Durme, Ting Qian, and Lenhart Schubert. 2008. Class-driven Attribute Extraction. In COLING. [find](https://scholar.google.com/scholar?q=Benjamin+Van+Durme%2C+Ting+Qian%2C+and+Lenhart+Schubert%2E+2008%2E+Class%2Ddriven+Attribute+Extraction%2E+In+COLING%2E)

*Updated 2 November 2026*