Adaptive Parser-Centric Text Normalization
Zhang, Congle and Baldwin, Tyler and Ho, Howard and Kimelfeld, Benny and Li, Yunyao

Article Structure

Abstract

Text normalization is an important first step towards enabling many Natural Language Processing (NLP) tasks over informal text.

Introduction

Text normalization is the task of transforming informal writing into its standard form in the language.

Related Work

Sproat et al.

Model

In this section we introduce our normalization framework, which draws inspiration from our previous work on spelling correction for search (Bao et al., 2011).

Instantiation

In this section, we discuss our instantiation of the model presented in the previous section.

Evaluation

In this section, we present an empirical study of our framework.

Discussion

The results presented in the previous section suggest that domain transfer using the proposed nor-

Conclusions

This work presents a framework for normalization with an eye towards domain adaptation.

Topics

gold standard

Appears in 10 sentences as: gold standard (12)
In Adaptive Parser-Centric Text Normalization
  1. (2) Update the weights by comparing the path generated in the previous step to the gold standard path.
    Page 5, “Model”
  2. First, we produce gold standard normalized data by manually normalizing sentences to their full grammatically correct form.
    Page 6, “Evaluation”
  3. In addition to the word-to-word mapping performed in typical normalization gold standard generation, this annotation procedure includes all actions necessary to make the sentence grammatical, such as word reordering, modifying capitalization, and removing emoticons.
    Page 6, “Evaluation”
  4. We then run an off-the-shelf dependency parser on the gold standard normalized data to produce our gold standard parses.
    Page 6, “Evaluation”
  5. To compare the parses produced over automatically normalized data to the gold standard , we look at the subjects, verbs, and objects (SVO) identified in each parse.
    Page 6, “Evaluation”
  6. GWZWN: The manual gold standard word-to-word normalizations of previous work (Choudhury et al., 2007; Han and Baldwin, 2011).
    Page 7, “Evaluation”
  7. Because the gold standard used in this work only provided word mappings for out-of-vocabulary words and did not enforce grammaticality, we reannotated the gold standard data2.
    Page 7, “Evaluation”
  8. Their original gold standard annotations were kept as a baseline.
    Page 7, “Evaluation”
  9. Most notably, both the generic and domain-specific systems outperformed the gold standard word-to-word normalizations.
    Page 7, “Evaluation”
  10. Perhaps most interestingly, the proposed approach performs on par with, and in several cases superior to, gold standard word-to-word annotations.
    Page 8, “Discussion”

See all papers in Proc. ACL 2013 that mention gold standard.

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

Back to top.

dependency parsing

Appears in 7 sentences as: dependency parser (3) dependency parsing (4)
In Adaptive Parser-Centric Text Normalization
  1. To address this problem, this work introduces an evaluation metric that ties normalization performance directly to the performance of a downstream dependency parser .
    Page 2, “Introduction”
  2. The goal is to evaluate the framework in two aspects: (1) usefulness for downstream applications (specifically dependency parsing ), and (2) domain adaptability.
    Page 6, “Evaluation”
  3. We then run an off-the-shelf dependency parser on the gold standard normalized data to produce our gold standard parses.
    Page 6, “Evaluation”
  4. These results validate the hypothesis that simple word-to-word normalization is insufficient if the goal of normalization is to improve dependency parsing ; even if a system could produce perfect word-to-word normalization, it would produce lower quality parses than those produced by our approach.
    Page 7, “Evaluation”
  5. The results in Section 5.2 establish a point that has often been assumed but, to the best of our knowledge, has never been explicitly shown: performing normalization is indeed beneficial to dependency parsing on informal text.
    Page 8, “Discussion”
  6. Additionally, this work introduces a parser-centric view of normalization, in which the performance of the normalizer is directly tied to the performance of a downstream dependency parser .
    Page 9, “Conclusions”
  7. Using this metric, this work established that, when dependency parsing is the goal, typical word-to-word normalization approaches are insufficient.
    Page 9, “Conclusions”

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

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

Back to top.

evaluation metric

Appears in 7 sentences as: evaluation metric (4) Evaluation Metrics (1) evaluation metrics (2)
In Adaptive Parser-Centric Text Normalization
  1. Another potential problem with state-of-the-art normalization is the lack of appropriate evaluation metrics .
    Page 1, “Introduction”
  2. For instance, it is unclear how performance measured by the typical normalization evaluation metrics of word error rate and BLEU score (Pap-ineni et al., 2002) translates into performance on a parsing task, where a well placed punctuation mark may provide more substantial improvements than changing a nonstandard word form.
    Page 2, “Introduction”
  3. To address this problem, this work introduces an evaluation metric that ties normalization performance directly to the performance of a downstream dependency parser.
    Page 2, “Introduction”
  4. In Section 5 we introduce the parser driven evaluation metric , and present experimental results of our model with respect to several baselines in three different domains.
    Page 2, “Introduction”
  5. 5.1 Evaluation Metrics
    Page 6, “Evaluation”
  6. Therefore, we propose a new evaluation metric that directly equates normalization performance with the performance of a common downstream application—dependency parsing.
    Page 6, “Evaluation”
  7. This evaluation metric allows for a deeper understanding of how certain normalization actions impact the output of the parser.
    Page 9, “Conclusions”

See all papers in Proc. ACL 2013 that mention evaluation metric.

See all papers in Proc. ACL that mention evaluation metric.

Back to top.

weight vector

Appears in 5 sentences as: weight vector (5)
In Adaptive Parser-Centric Text Normalization
  1. The conditional probability of an assignment 04, given an input sequence x and the weight vector 9 = (61, .
    Page 3, “Model”
  2. When performing inference, we wish to select the output sequence with the highest probability, given the input sequence X and the weight vector 9 (i.e., MAP inference).
    Page 4, “Model”
  3. A weight vector 9 = (61, .
    Page 4, “Model”
  4. path defined by 04, and Epmilxhefbj (04,-, Xi) is the expected value of that sum (over all legal assignments 04,-), assuming the current weight vector .
    Page 5, “Model”
  5. Instead of computing the expectation, we simply compute (DJ-(04f, Xi), where a: is the assignment with the highest probability, generated using the current weight vector .
    Page 5, “Model”

See all papers in Proc. ACL 2013 that mention weight vector.

See all papers in Proc. ACL that mention weight vector.

Back to top.

domain adaptability

Appears in 4 sentences as: domain adaptability (2) domain adaptation (2)
In Adaptive Parser-Centric Text Normalization
  1. Additionally, we design a cus-tomizable framework to address the often overlooked concept of domain adaptability , and illustrate that the system allows for transfer to new domains with a minimal amount of data and effort.
    Page 1, “Abstract”
  2. Similarly, our work is the first to prioritize domain adaptation during the new wave of text message normalization.
    Page 2, “Related Work”
  3. The goal is to evaluate the framework in two aspects: (1) usefulness for downstream applications (specifically dependency parsing), and (2) domain adaptability .
    Page 6, “Evaluation”
  4. This work presents a framework for normalization with an eye towards domain adaptation .
    Page 9, “Conclusions”

See all papers in Proc. ACL 2013 that mention domain adaptability.

See all papers in Proc. ACL that mention domain adaptability.

Back to top.

machine translation

Appears in 4 sentences as: machine translation (4)
In Adaptive Parser-Centric Text Normalization
  1. It is an important processing step for a wide range of Natural Language Processing (NLP) tasks such as text-to-speech synthesis, speech recognition, information extraction, parsing, and machine translation (Sproat et al., 2001).
    Page 1, “Introduction”
  2. Research on SMS and Twitter normalization has been roughly categorized as drawing inspiration from three other areas of NLP (Kobus et al., 2008): machine translation , spell checking, and automatic speech recognition.
    Page 2, “Related Work”
  3. The statistical machine translation (SMT) metaphor was the first proposed to handle the text normalization problem (Aw et al., 2006).
    Page 2, “Related Work”
  4. (2008) undertook a hybrid approach that pulls inspiration from both the machine translation and speech recognition metaphors.
    Page 2, “Related Work”

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

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

Back to top.

edit distance

Appears in 3 sentences as: edit distance (3)
In Adaptive Parser-Centric Text Normalization
  1. obtain a truth assignment 05°” from each yzgo'd by selecting an assignment 04 that minimizes the edit distance between ngId and the normalized
    Page 4, “Model”
  2. Here, y(a) denotes the normalized text implied by a, and DIST is a token-level edit distance .
    Page 4, “Model”
  3. Generator From To leave intact good good edit distance bac back lowercase NEED need capitalize it It Google spell disspaear disappear contraction wouldn’t would not slang language ima I am going to insert punctuation e .
    Page 5, “Instantiation”

See all papers in Proc. ACL 2013 that mention edit distance.

See all papers in Proc. ACL that mention edit distance.

Back to top.

part-of-speech

Appears in 3 sentences as: Part-of-speech (2) part-of-speech (3)
In Adaptive Parser-Centric Text Normalization
  1. Part-of-speech: Part-of-speech information can be used to produce features that encourage certain behavior, such as avoiding the deletion of noun phrases.
    Page 5, “Instantiation”
  2. We generate part-of-speech information over the original raw text using a Twitter part-of-speech tagger (Ritter et al., 2011).
    Page 5, “Instantiation”
  3. Of course, the part-of-speech information obtained this way is likely to be noisy, and we expect our learning algorithm to take that into account.
    Page 5, “Instantiation”

See all papers in Proc. ACL 2013 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