Learning to Paraphrase: An Unsupervised Approach Using Multiple-Sequence Alignment

By Regina Barzilay, Lillian Lee. North American Chapter of the Association for Computational Linguistics, 2003

538 citations recorded by the source corpus.

We address the text-to-text generation problem of sentence-level paraphrasing --- a phenomenon distinct from and more difficult than word- or phrase-level paraphrasing. Our approach applies multiple-sequence alignment to sentences gathered from unannotated comparable corpora: it learns a set of paraphrasing patterns represented by word lattice pairs and automatically determines how to apply these patterns to rewrite new sentences. The results of our evaluation experiments show that the system derives accurate paraphrases, outperforming baseline systems.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/10b8f21e57b3392ce623c374c2c039f811ce5f69

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Paraphrase-Driven Learning for Open Question Answering (2013, methodology): “The learning algorithms presented in this paper are similar to algorithms used for paraphrase extraction from sentence-aligned corpora (Barzi-lay and McKeown, 2001; Barzilay and Lee, 2003; Quirk et al., 2004; Bannard and Callison-Burch, 2005; Callison-Burch, 2008; Marton et al., 2009).”
Recognising Textual Entailment with Robust Logical Inference (2005, background): “Future work will be directed to the acquisition of targeted paraphrases (as for example in [ BL03 ]) that can be converted into background knowledge in the form of axioms.”
Generating Syntactically Controlled Paraphrases without Using Annotated Parallel Pairs (2021, methodology): “Traditional approaches usually require hand-crafted rules, such as rule-based methods (McKeown, 1983), thesaurus-based methods (Bolshakov and Gelbukh, 2004; Kauchak and Barzilay, 2006), and lattice matching methods (Barzilay and Lee, 2003).”
Using Dependency-Based Features to Take the ’Para-farce’ out of Paraphrase (2006, methodology): “The generation of paraphrases from raw text is a speciﬁc type of what is commonly referred to as text-to-text generation (Barzilay and Lee, 2003).”
Generating Phrasal and Sentential Paraphrases: A Survey of Data-Driven Methods (2010, methodology): “Once all the slotted lattices have been constructed for each corpus, Steps 19–24 try to match the slotted lattices extracted from one corpus to those extracted from the other by referring back to the sentence clusters from which the original lattices were Algorithm 4 (Barzilay and Lee 2003) .”
Monolingual Machine Translation for Paraphrase Generation (2004, methodology): “To evaluate the quality of generation, we followed the lead of Barzilay & Lee (2003).”
A Continuously Growing Dataset of Sentential Paraphrases (2017, background): “…GIZA++, Jacana and Sultan’s aligner outputs) and the PPDB (Ganitkevitch et al., 2013). lated versions of a foreign novel (Barzilay and Elhadad, 2003; Barzilay and Lee, 2003) or other texts (Cohn et al., 2008), multiple deﬁnitions of the same concept (Hashimoto et al., 2011), descriptions of the…”
CRUISE: Cold-Start New Skill Development via Iterative Utterance Generation (2018, background): “One class of approaches is to paraphrase user utterances to increase the number of training set (Barzilay and Lee, 2003; Quirk et al., 2004; Kauchak and Barzilay, 2006; Zhao et al., 2009; Prakash et al., 2016).”
Collecting Highly Parallel Data for Paraphrase Evaluation (2011, methodology): “Without these resources, researchers have resorted to developing their own small, ad hoc datasets (Barzilay and McKeown, 2001; Shinyama et al., 2002; Barzilay and Lee, 2003; Quirk et al., 2004; Dolan et al., 2004), and have often relied on human judgments to evaluate their results (Barzilay and…”
LexRank: Graph-based Centrality as Salience in Text Summarization (2004, background): “Many problems in NLP, e.g., parsing (Collins, 1997), word sense disambiguation (Yarowsky, 1995), and automatic paraphrasing (Barzilay & Lee, 2003) have benefited significantly by the introduction of robust statistical techniques.”
LCQMC:A Large-scale Chinese Question Matching Corpus (2018, background): “where Gn(S) is the set of n-grams in sentence S, and N is usually set to 4 (Barzilay and Lee, 2003).”
MUSS: Multilingual Unsupervised Sentence Simplification by Mining Paraphrases (2022, unknown): “Previous unsupervised paraphrasing research has aligned sentences from parallel corpora (Barzilay and Lee, 2003) with multiple objective functions (Liu et al.”
Paraphrase Generation: A Survey of the State of the Art (2021, background): “Due to the signiﬁcant manual efforts, some researchers have sought to collect paraphrase rules automatically (Lin and Pantel, 2001; Barzilay and Lee, 2003).”
Syntax-Guided Controlled Generation of Paraphrases (2020, methodology): “While generation of paraphrases has been addressed in the past using traditional methods (McKeown, 1983; Barzilay and Lee, 2003; Quirk et al., 2004; Hassan et al., 2007; Zhao et al., 2008; Madnani and Dorr, 2010; Wubben et al., 2010), they have recently been superseded by deep learning-based approac”
Unsupervised Construction of Large Paraphrase Corpora: Exploiting Massively Parallel News Sources (2004, background): “While the idea of exploiting multiple news reports for paraphrase acquisition is not new, previous efforts (for example, Shinyama et al. 2002; Barzilay and Lee 2003) have been restricted to at most two news sources.”
A Survey of Paraphrasing and Textual Entailment Methods (2009, methodology): “…different scores are obtained, depending on whether the extracted pairs are considered in particular contexts or not, and whether they are required to be interchangeable in grammatical sentences (Bannard & Callison-Burch, 2005; Barzilay & Lee, 2003; Callison-Burch, 2008; Zhao et al., 2008).”
Dynamic Pooling and Unfolding Recursive Autoencoders for Paraphrase Detection (2011, methodology): “Early approaches were based purely on lexical matching techniques [22, 23, 19, 24].”