A Comparison of Vector-based Representations for Semantic Composition

By William Blacoe, Mirella Lapata. Conference on Empirical Methods in Natural Language Processing, 2012

296 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/7988ef10dc9770e5fa4dc40d5d2f3693fd2ed917

How researchers see this paper

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

Citations in the corpus

Re-embedding words (2013, background): “Perhaps the most direct approach is to compute a weighted linear combination of the embeddings for words that appear in the document to be classiﬁed, as done in (Maas et al., 2011) and (Blacoe and Lapata, 2012).”
From Paraphrase Database to Compositional Paraphrase Model and Back (2015, methodology): “Simple but effective compositional models were studied by Mitchell and Lapata (2008; 2010) and Blacoe and Lapata (2012).”
Linguistic Regularities in Sparse and Explicit Word Representations (2014, background): “Blacoe and Lapata (2012) compare different arithmetic functions across multiple representations (including embeddings) on a range of compositionality benchmarks.”
Distributional Models of Word Meaning (2018, methodology): “Still, simple additive or multiplicative methods are very hard to beat by more complex distributional methods for semantic compositionality (Blacoe & Lapata 2012).”
A Tensor-based Factorization Model of Semantic Compositionality (2013, unknown): “However, the absolute gains of the systems remain a bit unclear, and a simple method of composition – vector multiplication – often seems to produce the best results (Blacoe and Lapata, 2012).”
SemEval-2015 Task 1: Paraphrase and Semantic Similarity in Twitter (PIT) (2015, methodology): “These can be categorized into supervised methods (Qiu et al., 2006; Wan et al., 2006; Das and Smith, 2009; Socher et al., 2011; Blacoe and Lapata, 2012; Mad-nani et al., 2012; Ji and Eisenstein, 2013) and unsupervised methods (Mihalcea et al., 2006; Rus et al., 2008; Fernando and Stevenson, 2008;…”
Compositional Operators in Distributional Semantics (2014, background): “In fact, the comparative study of Blacoe and Lapata (2012) suggests something really surprising: that, for certain tasks, additive and multiplicative models can be almost as much effective as state-of-the-art deep learning models, which will be the subject of Section 4.4.”
Monolingual and Cross-Lingual Information Retrieval Models Based on (Bilingual) Word Embeddings (2015, background): “5.3, we believe that better results may be achieved by introducing more complex composition models available in the literature (e.g., [4, 27]).”
Prior Disambiguation of Word Tensors for Constructing Sentence Vectors (2013, result): “Despite the obvious benefits of the tensor-based approaches, this work suggests for one more time that vector mixture models might constitute a hardto-beat baseline; similar observations have been made, for example, in the comparative study of Blacoe and Lapata (2012). However, when trying to interp”
The Role of Syntax in Vector Space Models of Compositional Semantics (2013, methodology): “Previously, RAE have successfully been applied to a number of tasks including sentiment analysis, paraphrase (Blacoe and Lapata, 2012; Socher et al., 2011b; Socher et al., 2012a).”
Revisiting Recurrent Networks for Paraphrastic Sentence Embeddings (2017, unknown): “Simple operations based on vector addition and multiplication typically serve as strong baselines (Mitchell and Lapata, 2008, 2010; Blacoe and Lapata, 2012).”
Multi-Perspective Sentence Similarity Modeling with Convolutional Neural Networks (2015, unknown): “9% Blacoe and Lapata (2012) 73.”
MultiGranCNN: An Architecture for General Matching of Text Chunks on Multiple Levels of Granularity (2015, background): “Most prior work (e.g., Blacoe and Lapata (2012; Hu et al. (2014)) has neglected the need for multi-granular comparability and performed matching within ﬁxed levels only, e.g., only words were matched with words or only sentences with sentences.”
Towards Universal Paraphrastic Sentence Embeddings (2015, background): “Follow-up work by Blacoe & Lapata (2012) found again that simple operations such as vector ad-2 dition performed strongly.”
On Deep Multi-View Representation Learning (2015, methodology): “We simply add the projections of the two words in each bigram to obtain an L -dimensional representation of the bigram, as done in prior work ( Blacoe and Lapata, 2012).”
Evaluating Neural Word Representations in Tensor-Based Compositional Settings (2014, methodology): “…level (Levy et al., 2014; Baroni et al., 2014), we are aware of only one work that attempts a comparison of the two approaches in compositional settings (Blacoe and Lapata, 2012), and this is limited to additive and multiplicative composition, compared against composition via a neural autoencoder.”
Multichannel Variable-Size Convolution for Sentence Classification (2015, background): “Blacoe and Lapata (2012) represented a sentence by element-wise addition, multiplication, or recursive autoencoder over embeddings of component single words.”
ABCNN: Attention-Based Convolutional Neural Network for Modeling Sentence Pairs (2015, methodology): “For TE, Bowman et al. (2015b) use recursive neural networks to encode entailment on SICK (Marelli et al., 2014b).”