We propose SentiBERT, a variant of BERT that effectively captures compositional sentiment semantics. The model incorporates contextualized representation with binary constituency parse tree to capture semantic composition. Comprehensive experiments demonstrate that SentiBERT achieves competitive performance on phrase-level sentiment classification. We further demonstrate that the sentiment composition learned from the phrase-level annotations on SST can be transferred to other sentiment analysis tasks as well as related tasks, such as emotion classification tasks. Moreover, we conduct ablation studies and design visualization methods to understand SentiBERT. We show that SentiBERT is better than baseline approaches in capturing negation and the contrastive relation and model the compositional sentiment semantics.
Functional Distributional Semantics provides a linguistically interpretable framework for distributional semantics, by representing the meaning of a word as a function (a binary classifier), instead of a vector. However, the large number of latent variables means that inference is computationally expensive, and training a model is therefore slow to converge. In this paper, I introduce the Pixie Autoencoder, which augments the generative model of Functional Distributional Semantics with a graph-convolutional neural network to perform amortised variational inference. This allows the model to be trained more effectively, achieving better results on two tasks (semantic similarity in context and semantic composition), and outperforming BERT, a large pre-trained language model.
The impressive performance of neural networks on natural language processing tasks attributes to their ability to model complicated word and phrase compositions. To explain how the model handles semantic compositions, we study hierarchical explanation of neural network predictions. We identify non-additivity and context independent importance attributions within hierarchies as two desirable properties for highlighting word and phrase compositions. We show some prior efforts on hierarchical explanations, e.g. contextual decomposition, do not satisfy the desired properties mathematically, leading to inconsistent explanation quality in different models. In this paper, we start by proposing a formal and general way to quantify the importance of each word and phrase. Following the formulation, we propose Sampling and Contextual Decomposition (SCD) algorithm and Sampling and Occlusion (SOC) algorithm. Human and metrics evaluation on both LSTM models and BERT Transformer models on multiple datasets show that our algorithms outperform prior hierarchical explanation algorithms. Our algorithms help to visualize semantic composition captured by models, extract classification rules and improve human trust of models. Project page: https://inklab.usc.edu/hiexpl/
In this paper we present OSCAR (Ontology-based Semantic Composition Augmented Regularization), a method for injecting task-agnostic knowledge from an Ontology or knowledge graph into a neural network during pretraining. We evaluated the impact of including OSCAR when pretraining BERT with Wikipedia articles by measuring the performance when fine-tuning on two question answering tasks involving world knowledge and causal reasoning and one requiring domain (healthcare) knowledge and obtained 33:3%, 18:6%, and 4% improved accuracy compared to pretraining BERT without OSCAR and obtaining new state-of-the-art results on two of the tasks.
Composition models of distributional semantics are used to construct phrase representations from the representations of their words. Composition models are typically situated on two ends of a spectrum. They either have a small number of parameters but compose all phrases in the same way, or they perform word-specific compositions at the cost of a far larger number of parameters. In this paper we propose transformation weighting (TransWeight), a composition model that consistently outperforms existing models on nominal compounds, adjective-noun phrases and adverb-adjective phrases in English, German and Dutch. TransWeight drastically reduces the number of parameters needed compared to the best model in the literature by composing similar words in the same way.
Capturing the meaning of sentences has long been a challenging task. Current models tend to apply linear combinations of word features to conduct semantic composition for bigger-granularity units e.g. phrases, sentences, and documents. However, the semantic linearity does not always hold in human language. For instance, the meaning of the phrase `ivory tower' can not be deduced by linearly combining the meanings of `ivory' and `tower'. To address this issue, we propose a new framework that models different levels of semantic units (e.g. sememe, word, sentence, and semantic abstraction) on a single \textit{Semantic Hilbert Space}, which naturally admits a non-linear semantic composition by means of a complex-valued vector word representation. An end-to-end neural network~\footnote{https://github.com/wabyking/qnn} is proposed to implement the framework in the text classification task, and evaluation results on six benchmarking text classification datasets demonstrate the effectiveness, robustness and self-explanation power of the proposed model. Furthermore, intuitive case studies are conducted to help end users to understand how the framework works.
In recent years, with the prevalence of social media and smart devices, people causally reveal their locations such as shops, hotels, and restaurants in their tweets. Recognizing and linking such fine-grained location mentions to well-defined location profiles are beneficial for retrieval and recommendation systems. In this paper, we propose DLocRL, a new deep learning pipeline for fine-grained location recognition and linking in tweets, and verify its effectiveness on a real-world Twitter dataset.
Named entity recognition (NER) is the task to identify mentions of rigid designators from text belonging to predefined semantic types such as person, location, organization etc. NER always serves as the foundation for many natural language applications such as question answering, text summarization, and machine translation. Early NER systems got a huge success in achieving good performance with the cost of human engineering in designing domain-specific features and rules. In recent years, deep learning, empowered by continuous real-valued vector representations and semantic composition through nonlinear processing, has been employed in NER systems, yielding stat-of-the-art performance. In this paper, we provide a comprehensive review on existing deep learning techniques for NER. We first introduce NER resources, including tagged NER corpora and off-the-shelf NER tools. Then, we systematically categorize existing works based on a taxonomy along three axes: distributed representations for input, context encoder, and tag decoder. Next, we survey the most representative methods for recent applied techniques of deep learning in new NER problem settings and applications. Finally, we present readers with the challenges faced by NER systems and outline future directions in this area.
Recursive Neural Network (RecNN), a type of models which compose words or phrases recursively over syntactic tree structures, has been proven to have superior ability to obtain sentence representation for a variety of NLP tasks. However, RecNN is born with a thorny problem that a shared compositional function for each node of trees can't capture the complex semantic compositionality so that the expressive power of model is limited. In this paper, in order to address this problem, we propose Tag-Guided HyperRecNN/TreeLSTM (TG-HRecNN/TreeLSTM), which introduces hypernetwork into RecNNs to take as inputs Part-of-Speech (POS) tags of word/phrase and generate the semantic composition parameters dynamically. Experimental results on five datasets for two typical NLP tasks show proposed models both obtain significant improvement compared with RecNN and TreeLSTM consistently. Our TG-HTreeLSTM outperforms all existing RecNN-based models and achieves or is competitive with state-of-the-art on four sentence classification benchmarks. The effectiveness of our models is also demonstrated by qualitative analysis.
Given appropriate representations of the semantic relations between carpenter and wood and between mason and stone (for example, vectors in a vector space model), a suitable algorithm should be able to recognize that these relations are highly similar (carpenter is to wood as mason is to stone; the relations are analogous). Likewise, with representations of dog, house, and kennel, an algorithm should be able to recognize that the semantic composition of dog and house, dog house, is highly similar to kennel (dog house and kennel are synonymous). It seems that these two tasks, recognizing relations and compositions, are closely connected. However, up to now, the best models for relations are significantly different from the best models for compositions. In this paper, we introduce a dual-space model that unifies these two tasks. This model matches the performance of the best previous models for relations and compositions. The dual-space model consists of a space for measuring domain similarity and a space for measuring function similarity. Carpenter and wood share the same domain, the domain of carpentry. Mason and stone share the same domain, the domain of masonry. Carpenter and mason share the same function, the function of artisans. Wood and stone share the same function, the function of materials. In the composition dog house, kennel has some domain overlap with both dog and house (the domains of pets and buildings). The function of kennel is similar to the function of house (the function of shelters). By combining domain and function similarities in various ways, we can model relations, compositions, and other aspects of semantics.
Semantic composition is the task of understanding the meaning of text by composing the meanings of the individual words in the text. Semantic decomposition is the task of understanding the meaning of an individual word by decomposing it into various aspects (factors, constituents, components) that are latent in the meaning of the word. We take a distributional approach to semantics, in which a word is represented by a context vector. Much recent work has considered the problem of recognizing compositions and decompositions, but we tackle the more difficult generation problem. For simplicity, we focus on noun-modifier bigrams and noun unigrams. A test for semantic composition is, given context vectors for the noun and modifier in a noun-modifier bigram ("red salmon"), generate a noun unigram that is synonymous with the given bigram ("sockeye"). A test for semantic decomposition is, given a context vector for a noun unigram ("snifter"), generate a noun-modifier bigram that is synonymous with the given unigram ("brandy glass"). With a vocabulary of about 73,000 unigrams from WordNet, there are 73,000 candidate unigram compositions for a bigram and 5,300,000,000 (73,000 squared) candidate bigram decompositions for a unigram. We generate ranked lists of potential solutions in two passes. A fast unsupervised learning algorithm generates an initial list of candidates and then a slower supervised learning algorithm refines the list. We evaluate the candidate solutions by comparing them to WordNet synonym sets. For decomposition (unigram to bigram), the top 100 most highly ranked bigrams include a WordNet synonym of the given unigram 50.7% of the time. For composition (bigram to unigram), the top 100 most highly ranked unigrams include a WordNet synonym of the given bigram 77.8% of the time.
The chain-structured long short-term memory (LSTM) has showed to be effective in a wide range of problems such as speech recognition and machine translation. In this paper, we propose to extend it to tree structures, in which a memory cell can reflect the history memories of multiple child cells or multiple descendant cells in a recursive process. We call the model S-LSTM, which provides a principled way of considering long-distance interaction over hierarchies, e.g., language or image parse structures. We leverage the models for semantic composition to understand the meaning of text, a fundamental problem in natural language understanding, and show that it outperforms a state-of-the-art recursive model by replacing its composition layers with the S-LSTM memory blocks. We also show that utilizing the given structures is helpful in achieving a performance better than that without considering the structures.
Representation learning of knowledge bases (KBs) aims to embed both entities and relations into a low-dimensional space. Most existing methods only consider direct relations in representation learning. We argue that multiple-step relation paths also contain rich inference patterns between entities, and propose a path-based representation learning model. This model considers relation paths as translations between entities for representation learning, and addresses two key challenges: (1) Since not all relation paths are reliable, we design a path-constraint resource allocation algorithm to measure the reliability of relation paths. (2) We represent relation paths via semantic composition of relation embeddings. Experimental results on real-world datasets show that, as compared with baselines, our model achieves significant and consistent improvements on knowledge base completion and relation extraction from text.
Classical scope-assignment strategies for multi-quantifier sentences involve quantifier phrase (QP)-movement. More recent continuation-based approaches provide a compelling alternative, for they interpret QP's in situ - without resorting to Logical Forms or any structures beyond the overt syntax. The continuation-based strategies can be divided into two groups: those that locate the source of scope-ambiguity in the rules of semantic composition and those that attribute it to the lexical entries for the quantifier words. In this paper, we focus on the former operation-based approaches and the nature of the semantic operations involved. More specifically, we discuss three such possible operation-based strategies for multi-quantifier sentences, together with their relative merits and costs.
Distributional models are derived from co-occurrences in a corpus, where only a small proportion of all possible plausible co-occurrences will be observed. This results in a very sparse vector space, requiring a mechanism for inferring missing knowledge. Most methods face this challenge in ways that render the resulting word representations uninterpretable, with the consequence that semantic composition becomes hard to model. In this paper we explore an alternative which involves explicitly inferring unobserved co-occurrences using the distributional neighbourhood. We show that distributional inference improves sparse word representations on several word similarity benchmarks and demonstrate that our model is competitive with the state-of-the-art for adjective-noun, noun-noun and verb-object compositions while being fully interpretable.
A Semantic Compositional Network (SCN) is developed for image captioning, in which semantic concepts (i.e., tags) are detected from the image, and the probability of each tag is used to compose the parameters in a long short-term memory (LSTM) network. The SCN extends each weight matrix of the LSTM to an ensemble of tag-dependent weight matrices. The degree to which each member of the ensemble is used to generate an image caption is tied to the image-dependent probability of the corresponding tag. In addition to captioning images, we also extend the SCN to generate captions for video clips. We qualitatively analyze semantic composition in SCNs, and quantitatively evaluate the algorithm on three benchmark datasets: COCO, Flickr30k, and Youtube2Text. Experimental results show that the proposed method significantly outperforms prior state-of-the-art approaches, across multiple evaluation metrics.
Count-based distributional semantic models suffer from sparsity due to unobserved but plausible co-occurrences in any text collection. This problem is amplified for models like Anchored Packed Trees (APTs), that take the grammatical type of a co-occurrence into account. We therefore introduce a novel form of distributional inference that exploits the rich type structure in APTs and infers missing data by the same mechanism that is used for semantic composition.
Semantic composition remains an open problem for vector space models of semantics. In this paper, we explain how the probabilistic graphical model used in the framework of Functional Distributional Semantics can be interpreted as a probabilistic version of model theory. Building on this, we explain how various semantic phenomena can be recast in terms of conditional probabilities in the graphical model. This connection between formal semantics and machine learning is helpful in both directions: it gives us an explicit mechanism for modelling context-dependent meanings (a challenge for formal semantics), and also gives us well-motivated techniques for composing distributed representations (a challenge for distributional semantics). We present results on two datasets that go beyond word similarity, showing how these semantically-motivated techniques improve on the performance of vector models.
Tree-structured neural network architectures for sentence encoding draw inspiration from the approach to semantic composition generally seen in formal linguistics, and have shown empirical improvements over comparable sequence models by doing so. Moreover, adding multiplicative interaction terms to the composition functions in these models can yield significant further improvements. However, existing compositional approaches that adopt such a powerful composition function scale poorly, with parameter counts exploding as model dimension or vocabulary size grows. We introduce the Lifted Matrix-Space model, which uses a global transformation to map vector word embeddings to matrices, which can then be composed via an operation based on matrix-matrix multiplication. Its composition function effectively transmits a larger number of activations across layers with relatively few model parameters. We evaluate our model on the Stanford NLI corpus, the Multi-Genre NLI corpus, and the Stanford Sentiment Treebank and find that it consistently outperforms TreeLSTM (Tai et al., 2015), the previous best known composition function for tree-structured models.
Semantic composition functions have been playing a pivotal role in neural representation learning of text sequences. In spite of their success, most existing models suffer from the underfitting problem: they use the same shared compositional function on all the positions in the sequence, thereby lacking expressive power due to incapacity to capture the richness of compositionality. Besides, the composition functions of different tasks are independent and learned from scratch. In this paper, we propose a new sharing scheme of composition function across multiple tasks. Specifically, we use a shared meta-network to capture the meta-knowledge of semantic composition and generate the parameters of the task-specific semantic composition models. We conduct extensive experiments on two types of tasks, text classification and sequence tagging, which demonstrate the benefits of our approach. Besides, we show that the shared meta-knowledge learned by our proposed model can be regarded as off-the-shelf knowledge and easily transferred to new tasks.