A Deep Architecture for Matching Short Texts

By Zhengdong Lu, Hang Li. 2013

282 citations recorded by the source corpus.

How researchers see this paper

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

In Natural Language Processing

Citations in the corpus

A Deep Relevance Matching Model for Ad-hoc Retrieval (2016, methodology): “Although various deep matching models have been proposed under such a general matching problem formalization, most of them have only been demonstrated to be eﬀective on a set of NLP tasks such as paraphrase identiﬁcation and QA [11, 26].”
Learning Deep Transformer Models for Machine Translation (2019, neutral): “In contrast, our approach allows the encoder layers to interact earlier, which has been proven to be effective in machine translation (He et al., 2018) and text match (Lu and Li, 2013).”
Neural Approaches to Conversational AI (2018, neutral): “ng new sentences word-by-word), a more conservative solution to mitigating blandness is to replace generation-based models with retrieval-based models for response generation (Jafarpour et al., 2010; Lu and Li, 2014; Inaba and Takahashi, 2016; Al-Rfou et al., 2016; Yan et al., 2016), in which the po”
Convolutional Neural Network Architectures for Matching Natural Language Sentences (2014, methodology): “…of two short-texts are calculated with an MLP with the embedding of the two documents as input; • D EEP M ATCH : We take the matching model in [13] and train it on our datasets with 3 hidden layers and 1,000 hidden nodes in the ﬁrst hidden layer; • U RAE+MLP: We use the Unfolding Recursive…”
Convolutional Neural Network for Paraphrase Identification (2015, background): “Lu and Li (2013) developed a deep NN to match short texts, where interactions between components within the two objects were considered.”
Chat as Expected: Learning to Manipulate Black-box Neural Dialogue Models (2020, neutral): “, 1996), retrieval-based methods (Lu and Li, 2013; Hu et al., 2014) to learning-based methods (Shang et al.”
Challenges in Building Intelligent Open-domain Dialog Systems (2019, methodology): “e top-scored candidate as output response Y using the matching function Pθ(Y|X,C), which can be implemented using either traditional learning-to-rank algorithms [68], or modern neural matching models [25, 46, 73]. The model parameters θis commonly 7Hereafter, we will use X ⊕C to denote the input que”
Learning to Match using Local and Distributed Representations of Text for Web Search (2016, background): “[12] classify recent DNN models for short-text matching as either interaction-focused [15, 22, 31] or representation-focused [15, 16, 35–37].”
From Eliza to XiaoIce: challenges and opportunities with social chatbots (2018, methodology): “generated by two approaches: retrieval-based (Lu and Li, 2013; Li et al., 2016; Yan et al., 2016) or generation-based (Vinyals and Le, 2015; Sordoni”
A Survey on Dialogue Systems: Recent Advances and New Frontiers (2017, methodology): “onds to the response vector representation, the matching function between xand ycan be as simply as bilinear matching: match(x,y) = xTAy, where A is a pre-determined matrix, or more complicated ones. [49] proposed a DNN-based matching model for short textsresponse selection and combined the localnes”
Modeling Relational Information in Question-Answer Pairs with Convolutional Neural Networks (2016, methodology): “…been successfully applied to various sentence classiﬁcation tasks, e.g., (Kalchbrenner et al., 2014; Kim, 2014), and for modelling text pairs, e.g. (Lu and Li, 2013; Hu et al., 2014), where in the latter model they use up to 3 convolution-pooling layers, while in our experiments deeper…”
A Short Texts Matching Method Using Shallow Features and Deep Features (2014, methodology): “Lu and Li [16] use a deep architecture to combine the localness and hierarchy intrinsic to natural language problem by using a massive dataset.”
Modelling input texts: from Tree Kernels to Deep Learning (2015, unknown): “[60, 84].”
Deep Learning for Biomedical Information Retrieval: Learning Textual Relevance from Click Logs (2017, unknown): “Examples are described in (Hu et al., 2014; Lu and Li, 2013).”
Convolutional Neural Networks vs. Convolution Kernels: Feature Engineering for Answer Sentence Reranking (2016, methodology): “Recently, deep learning approaches have been successfully applied to various sentence classiﬁcation tasks, e.g., (Kalchbrenner et al., 2014; Kim, 2014), and for automatically modeling text pairs, e.g., (Lu and Li, 2013; Hu et al., 2014).”
From Neural Re-Ranking to Neural Ranking: Learning a Sparse Representation for Inverted Indexing (2018, unknown): “As another example, DeepMatch [27] computes the matching score for a text pair by consideringword sequences.”
A Latent Semantic Model with Convolutional-Pooling Structure for Information Retrieval (2014, unknown): “Recently, a neural network based DeepMatch model is also proposed to directly capture the correspondence between two short texts without explicitly relying on semantic vector representations [23].”