No file available
To read the file of this research,
you can request a copy directly from the authors.
Using k-way Co-occurrences for Learning Word Embeddings
Preprints and early-stage research may not have been peer reviewed yet.
Abstract
Co-occurrences between two words provide useful insights into the semantics of those words. Consequently, numerous prior work on word embedding learning have used co-occurrences between two words as the training signal for learning word embeddings. However, in natural language texts it is common for multiple words to be related and co-occurring in the same context. We extend the notion of co-occurrences to cover -way co-occurrences among a set of k-words. Specifically, we prove a theoretical relationship between the joint probability of words, and the sum of norms of their embeddings. Next, we propose a learning objective motivated by our theoretical result that utilises k-way co-occurrences for learning word embeddings. Our experimental results show that the derived theoretical relationship does indeed hold empirically, and despite data sparsity, for some smaller k values, k-way embeddings perform comparably or better than 2-way embeddings in a range of tasks.
ResearchGate has not been able to resolve any citations for this publication.
We investigate the task of assessing sentence-level prompt relevance in learner essays. Various systems using word overlap, neural embeddings and neural compositional models are evaluated on two datasets of learner writing. We propose a new method for sentence-level similarity calculation, which learns to adjust the weights of pre-trained word embeddings for a specific task, achieving substantially higher accuracy compared to other relevant baselines.
We present the Siamese Continuous Bag of Words (Siamese CBOW) model, a neural network for efficient estimation of high-quality sentence embeddings. Averaging the embeddings of words in a sentence has proven to be a surprisingly successful and efficient way of obtaining sentence embeddings. However, word embeddings trained with the methods currently available are not optimized for the task of sentence representation, and, thus, likely to be suboptimal. Siamese CBOW handles this problem by training word embeddings directly for the purpose of being averaged. The underlying neural network learns word embeddings by predicting, from a sentence representation, its surrounding sentences. We show the robustness of the Siamese CBOW model by evaluating it on 20 datasets stemming from a wide variety of sources.
We present two simple modifications to the models in the popular Word2Vec tool, in order to generate embeddings more suited to tasks involving syntax. The main issue with the original models is the fact that they are insensitive to word order. While order independence is useful for inducing semantic representations, this leads to suboptimal results when they are used to solve syntax-based problems. We show improvements in part-of-speech tagging and dependency parsing using our proposed models.
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.
We present SimLex-999, a gold standard resource for evaluating distributional semantic models that improves on existing resources in several important ways. First, in contrast to gold standards such as WordSim-353 and MEN, it explicitly quantifies similarity rather than association or relatedness so that pairs of entities that are associated but not actually similar (Freud, psychology) have a low rating. We show that, via this focus on similarity, SimLex-999 incentivizes the development of models with a different, and arguably wider, range of applications than those which reflect conceptual association. Second, SimLex-999 contains a range of concrete and abstract adjective, noun, and verb pairs, together with an independent rating of concreteness and (free) association strength for each pair. This diversity enables fine-grained analyses of the performance of models on concepts of different types, and consequently greater insight into how architectures can be improved. Further, unlike existing gold standard evaluations, for which automatic approaches have reached or surpassed the inter-annotator agreement ceiling, state-of-the-art models perform well below this ceiling on SimLex-999. There is therefore plenty of scope for SimLex-999 to quantify future improvements to distributional semantic models, guiding the development of the next generation of representation-learning architectures.
Ordering information is a difficult but an important task for applications generating natural-language texts such as multi-document summarization, question answering, and concept-to-text generation. In multi-document summarization, information is selected from a set of source documents. Therefore, the optimal ordering of those selected pieces of information to create a coherent summary is not obvious. Improper ordering of information in a summary can both confuse the reader and deteriorate the readability of the summary. Therefore, it is vital to properly order the information in multi-document summarization. We model the problem of sentence ordering in multi-document summarization as a one of learning the optimal combination of preference experts that determine the ordering between two given sentences. To capture the preference of a sentence against another sentence, we define five preference experts: chronology, probabilistic, topical-closeness, precedence, and succession. We use summaries ordered by human annotators as training data to learn the optimal combination of the different preference experts. Finally, the learnt combination is applied to order sentences extracted in a multi-document summarization system. The proposed sentence ordering algorithm considers pairwise comparisons between sentences to determine a total ordering, using a greedy search algorithm, thereby avoiding the combinatorial time complexity typically associated with total ordering tasks. This enables us to efficiently order sentences in longer summaries, thereby rendering the proposed approach useable in real-world text summarization systems. We evaluate the sentence orderings produced by the proposed method and numerous other baselines using both semi-automatic evaluation measures as well as performing a subjective evaluation.
Up to now, work on semantic relations has fo-cused on relation classification: recognizing whether a given instance (a word pair such as virus:flu) belongs to a specific relation class (such as CAUSE:EFFECT). However, instances of a single relation class may still have signif-icant variability in how characteristic they are of that class. We present a new SemEval task based on identifying the degree of prototypi-cality for instances within a given class. As a part of the task, we have assembled the first dataset of graded relational similarity ratings across 79 relation categories. Three teams submitted six systems, which were evaluated using two methods.
The performance of Differential Evolution (DE) algorithm is significantly affected by its parameter setting. But the choice of parameters is heavily dependent on the problem characteristics. Therefore, recently a couple of adaptation schemes that automatically adjust DE parameters have been proposed. The current work presents another adaptation scheme for DE parameters namely amplification factor and crossover rate. We systematically analyze the effectiveness of the proposed adaptation scheme for DE parameters using a standard benchmark suite consisting of ten functions. The undertaken empirical study shows that the proposed adaptive DE (aDE) algorithm exhibits an overall better performance compared to other prominent adaptive DE algorithms as well as canonical DE.
Merchants selling products on the Web often ask their customers to review the products that they have purchased and the associated services. As e-commerce is becoming more and more popular, the number of customer reviews that a product receives grows rapidly. For a popular product, the number of reviews can be in hundreds or even thousands. This makes it difficult for a potential customer to read them to make an informed decision on whether to purchase the product. It also makes it difficult for the manufacturer of the product to keep track and to manage customer opinions. For the manufacturer, there are additional difficulties because many merchant sites may sell the same product and the manufacturer normally produces many kinds of products. In this research, we aim to mine and to summarize all the customer reviews of a product. This summarization task is different from traditional text summarization because we only mine the features of the product on which the customers have expressed their opinions and whether the opinions are positive or negative. We do not summarize the reviews by selecting a subset or rewrite some of the original sentences from the reviews to capture the main points as in the classic text summarization. Our task is performed in three steps: (1) mining product features that have been commented on by customers; (2) identifying opinion sentences in each review and deciding whether each opinion sentence is positive or negative; (3) summarizing the results. This paper proposes several novel techniques to perform these tasks. Our experimental results using reviews of a number of products sold online demonstrate the effectiveness of the techniques.
Identifying discourse relations in a text is essential for various tasks in Natural Language Processing, such as automatic text summa- rization, question-answering, and dialogue generation. The first step of this process is segmenting a text into elementary units. In this paper, we present a novel model of discourse segmentation based on sequential data labeling. Namely, we use Conditional Random Fields to train a dis- course segmenter on the RST Discourse Treebank, using a set of lexical and syntactic features. Our system is compared to other statistical and rule-based segmenters, including one based on Support Vector Machines. Experimental results indicate that our sequential model outperforms cur- rent state-of-the-art discourse segmenters, with an F-score of 0.94. This performance level is close to the human agreement F-score of 0.98.
Latent relational search (LRS) is a novel approach for mapping knowledge across two domains. Given a source domain knowledge concerning the Moon, "The Moon is a satellite of the Earth", one can form a question {(Moon, Earth), (Ganymede, ?)} to query an LRS engine for new knowledge in the target domain concerning the Ganymede. An LRS engine relies on some supporting sentences such as "Ganymede is a natural satellite of Jupiter." to retrieve and rank "Jupiter" as the first answer. This paper proposes cross-language latent relational search (CLRS) to extend the knowledge mapping capability of LRS from cross-domain knowledge mapping to cross-domain and cross-language knowledge mapping. In CLRS, the supporting sentences for the source pair might be in a different language with that of the target pair. We represent the relation between two entities in an entity pair by lexical patterns of the context surrounding the two entities. We then propose a novel hybrid lexical pattern clustering algorithm to capture the semantic similarity between paraphrased lexical patterns across languages. Experiments on Japanese-English datasets show that the proposed method achieves an MRR of 0.579 for CLRS task, which is comparable to the MRR of an existing monolingual LRS engine. Copyright © 2011, Association for the Advancement of Artificial Intelligence. All rights reserved.
Ordering information is a difficult but a important task for natural language generation applications. A wrong order of information not only makes it difficult to understand, but also conveys an entirely different idea to the reader. This paper proposes an algorithm that learns orderings from a set of human ordered texts. Our model consists of a set of ordering experts. Each expert gives its precedence preference between two sentences. We combine these preferences and order sentences. We also propose two new metrics for the evaluation of sentence orderings. Our experimental results show that the proposed algorithm outperforms the existing methods in all evaluation metrics.
Semantic similarity measures play important roles in infor- mation retrieval and Natural Language Processing. Previ- ous work in semantic web-related applications such as com- munity mining, relation extraction, automatic meta data extraction have used various semantic similarity measures. Despite the usefulness of semantic similarity measures in these applications, robustly measuring semantic similarity between two words (or entities) remains a challenging task. We propose a robust semantic similarity measure that uses the information available on the Web to measure similarity between words or entities. The proposed method exploits page counts and text snippets returned by a Web search engine. We deflne various similarity scores for two given words P and Q, using the page counts for the queries P, Q and P AND Q. Moreover, we propose a novel approach to compute semantic similarity using automatically extracted lexico-syntactic patterns from text snippets. These difierent similarity scores are integrated using support vector ma- chines, to leverage a robust semantic similarity measure. Experimental results on Miller-Charles benchmark dataset show that the proposed measure outperforms all the existing web-based semantic similarity measures by a wide margin, achieving a correlation coe-cient of 0:834. Moreover, the proposed semantic similarity measure signiflcantly improves the accuracy (F-measure of 0:78) in a community mining task, and in an entity disambiguation task, thereby verifying the capability of the proposed measure to capture semantic similarity using web content.
Several recent discourse parsers have employed fully-supervised machine learning approaches. These methods require human annotators to beforehand create an extensive training corpus, which is a time-consuming and costly process. On the other hand, unlabeled data is abundant and cheap to collect. In this paper, we propose a novel semi-supervised method for discourse relation classification based on the analysis of co-occurring features in unlabeled data, which is then taken into account for extending the feature vectors given to a classifier. Our experimental results on the RST Discourse Tree-bank corpus and Penn Discourse Treebank indicate that the proposed method brings a significant improvement in classification accuracy and macro-average F-score when small training datasets are used. For instance, with training sets of c.a. 1000 labeled instances, the proposed method brings improvements in accuracy and macro-average F-score up to 50% compared to a baseline classifier. We believe that the proposed method is a first step towards detecting low-occurrence relations, which is useful for domains with a lack of annotated data.
Researchers in both machine translation (e.g., Brown ., 1990) and bilingual lexicography (e.g., Klavans and Tzoukermann, 1990) have recently become interested in studying parallel texts, texts such as the Canadian Hansards (parliamentary proceedings) which are available in multiple languages (French and English). This paper describes a method for aligning sentences in these parallel texts, based on a simple statistical model of character lengths. The method was developed and tested on a small trilingual sample of Swiss economic reports. A much larger sample of 90 million words of Canadian Hansards has been aligned and donated to the ACL/DCI.
Measuring relational similarity between words is impor- tant in numerous natural language processing tasks such as solving analogy questions and classifying noun-modifier relations. We pro- pose a method to measure the similarity between semantic relations that hold between two pairs of words using a web search engine. First, each pair of words is represented by a vector of automatically extracted lexical patterns. Then a Support Vector Machine is trained to recognize word pairs with similar semantic relations. We evaluate the proposed method on SAT multiple-choice word-analogy ques- tions. The proposed method achieves a score of 40% which is com- parable with relational similarity measures which use manually cre- ated resources such as WordNet. The proposed method significantly reduces the time taken by previously proposed computationally in- tensive methods, such as latent relational analysis, to process 374 analogy questions from 8 days to less than 6 hours.
There are at least two kinds of similarity. Relational similarity is correspondence between relations, in contrast with attributional similarity, which is correspondence between attributes. When two words have a high degree of attributional similarity, we call them synonyms. When two pairs of words have a high degree of relational similarity, we say that their relations are analogous. For example, the word pair mason:stone is analogous to the pair carpenter:wood. This article introduces Latent Relational Analysis (LRA), a method for measuring relational similarity. LRA has potential applications in many areas, including information extraction, word sense disambiguation, and information retrieval. Recently the Vector Space Model (VSM) of information retrieval has been adapted to measuring relational similarity, achieving a score of 47% on a collection of 374 college-level multiple-choice word analogy questions. In the VSM approach, the relation between a pair of words is characterized by a vector of frequencies of predefined patterns in a large corpus. LRA extends the VSM approach in three ways: (1) The patterns are derived automatically from the corpus, (2) the Singular Value Decomposition (SVD) is used to smooth the frequency data, and (3) automatically generated synonyms are used to explore variations of the word pairs. LRA achieves 56% on the 374 analogy questions, statistically equivalent to the average human score of 57%. On the related problem of classifying semantic relations, LRA achieves similar gains over the VSM.
We present a new family of subgradient methods that dynamically incorporate knowledge of the geometry of the data observed in earlier iterations to perfo rm more informative gradient-based learning. The adaptation, in essence, allows us to find needl es in haystacks in the form of very predictive yet rarely observed features. Our paradigm stems from recent advances in online learning which employ proximal functions to control the gradient steps of the algorithm. We describe and analyze an apparatus for adaptively modifying the proximal function, which significantly simplifies the task of setting a learning rate and results in regret guar antees that are provably as good as the best proximal function that can be chosen in hindsight. We corroborate our theoretical results with experiments on a text classification task, showing substant ial improvements for classification with sparse datasets.
Computers understand very little of the meaning of human language. This profoundly limits our ability to give instructions to computers, the ability of computers to explain their actions to us, and the ability of computers to analyse and process text. Vector space models (VSMs) of semantics are beginning to address these limits. This paper surveys the use of VSMs for semantic processing of text. We organize the literature on VSMs according to the structure of the matrix in a VSM. There are currently three broad classes of VSMs, based on term-document, word-context, and pair-pattern matrices, yielding three classes of applications. We survey a broad range of applications in these three categories and we take a detailed look at a specific open source project in each category. Our goal in this survey is to show the breadth of applications of VSMs for semantics, to provide a new perspective on VSMs for those who are already familiar with the area, and to provide pointers into the literature for those who are less familiar with the field.
Much work has been done on the statistical analysis of text. In some cases reported in the literature, inappropriate statistical methods have been used, and statistical significance of results have not been addressed. In particular, asymptotic normality assumptions have often been used unjustifiably, leading to flawed results. This assumption of normal distribution limits the ability to analyze rare events. Unfortunately rare events do make up a large fraction of real text.
However, more applicable methods based on likelihood ratio tests are available that yield good results with relatively small samples. These tests can be implemented efficiently, and have been used for the detection of composite terms and for the determination of domain-specific terms. In some cases, these measures perform much better than the methods previously used. In cases where traditional contingency table methods work well, the likelihood ratio tests described here are nearly identical.
This paper describes the basis of a measure based on likelihood ratios that can be applied to the analysis of text.
This paper will propose an objective measure based on the information theoretic notion of mutual information, for estimating word association norms from computer readable corpora. (The standard method of obtaining word association norms, testing a few thousand subjects on a few hundred words, is both costly and unreliable.) The proposed measure, the association ratio, estimates word association norms directly from computer readable corpora, making it possible to estimate norms for tens of thousands of words
Semantic word embeddings represent the meaning of a word via a vector, and are created by diverse methods. Many use nonlinear operations on co-occurrence statistics, and have hand-tuned hyperparameters and reweighting methods.
This paper proposes a new generative model, a dynamic version of the log-linear topic model of Mnih and Hinton (2007). The methodological novelty is to use the prior to compute closed form expressions for word statistics. This provides a theoretical justification for nonlinear models like PMI, word2vec, and GloVe, as well as some hyperparameter choices. It also helps explain why low-dimensional semantic embeddings contain linear algebraic structure that allows solution of word analogies, as shown by Mikolov et al. (2013a) and many subsequent papers.
Experimental support is provided for the generative model assumptions, the most important of which is that latent word vectors are fairly uniformly dispersed in space.
Continuous word representations have been remarkably useful across NLP tasks but remain poorly understood. We ground word embeddings in semantic spaces studied in the cognitive-psychometric literature, taking these spaces as the primary objects to recover. To this end, we relate log co-occurrences of words in large corpora to semantic similarity assessments and show that co-occurrences are indeed consistent with an Euclidean semantic space hypothesis. Framing word embedding as metric recovery of a semantic space unifies existing word embedding algorithms, ties them to manifold learning, and demonstrates that existing algorithms are consistent metric recovery methods given co-occurrence counts from random walks. Furthermore, we propose a simple, principled, direct metric recovery algorithm that performs on par with the state-of-the-art word embedding and manifold learning methods. Finally, we complement recent focus on analogies by constructing two new inductive reasoning datasets—series completion and classification—and demonstrate that word embeddings can be used to solve them as well.
We propose two novel model architectures for computing continuous vector representations of words from very large data sets. The quality of these representations is measured in a word similarity task, and the results are compared to the previously best performing techniques based on different types of neural networks. We observe large improvements in accuracy at much lower computational cost, i.e. it takes less than a day to learn high quality word vectors from a 1.6 billion words data set. Furthermore, we show that these vectors provide state-of-the-art performance on our test set for measuring syntactic and semantic word similarities.
A common problem in knowledge representation and related fields is reasoning over a large joint knowledge graph, represented as triples of a relation between two entities. The goal of this paper is to develop a more powerful neural network model suitable for inference over these relationships. Previous models suffer from weak interaction between entities or simple linear projection of the vector space. We address these problems by introducing a neural tensor network (NTN) model which allow the entities and relations to interact multiplicatively. Additionally, we observe that such knowledge base models can be further improved by representing each entity as the average of vectors for the words in the entity name, giving an additional dimension of similarity by which entities can share statistical strength. We assess the model by considering the problem of predicting additional true relations between entities given a partial knowledge base. Our model outperforms previous models and can classify unseen relationships in WordNet and FreeBase with an accuracy of 86.2% and 90.0%, respectively.
Unsupervised word representations are very useful in NLP tasks both as inputs to learning algorithms and as extra word features in NLP systems. However, most of these models are built with only local context and one representation per word. This is problematic because words are often polysemous and global context can also provide useful information for learning word meanings. We present a new neural network architecture which 1) learns word embeddings that better capture the semantics of words by incorporating both local and global document context, and 2) accounts for homonymy and polysemy by learning multiple embeddings per word. We introduce a new dataset with human judgments on pairs of words in sentential context, and evaluate our model on it, showing that our model outperforms competitive baselines and other neural language models.
Our research aims at building computational models of word meaning that are perceptually grounded. Using computer vision techniques, we build visual and multimodal distributional models and compare them to standard textual models. Our results show that, while visual models with state-of-the-art computer vision techniques perform worse than textual models in general tasks (accounting for semantic relatedness), they are as good or better models of the meaning of words with visual correlates such as color terms, even in a nontrivial task that involves nonliteral uses of such words. Moreover, we show that visual and textual information are tapping on different aspects of meaning, and indeed combining them in multimodal models often improves performance.
The relationship between semantic and contextual similarity is investigated for pairs of nouns that vary from high to low semantic similarity. Semantic similarity is estimated by subjective ratings; contextual similarity is estimated by the method of sorting sentential contexts. The results show an inverse linear relationship between similarity of meaning and the discriminability of contexts. This relation, is obtained for two separate corpora of sentence contexts. It is concluded that, on average, for words in the same language drawn from the same syntactic and semantic categories, the more often two words can be substituted into the same contexts the more similar in meaning they are judged to be.
Neural probabilistic language models (NPLMs) have been shown to be competi- tive with and occasionally superior to the widely-used n-gram language models. The main drawback of NPLMs is their extremely long training and testing times. Morin and Bengio have proposed a hierarchical language model built around a binary tree of words, which was two orders of magnitude faster than the non- hierarchical model it was based on. However, it performed considerably worse than its non-hierarchical counterpart in spite of using a wo rd tree created using expert knowledge. We introduce a fast hierarchical language model along with a simple feature-based algorithm for automatic construction of word trees from the data. We then show that the resulting models can outperform non-hierarchical neural models as well as the best n-gram models.
We describe a single convolutional neural net- work architecture that, given a sentence, out- puts a host of language processing predic- tions: part-of-speech tags, chunks, named en- tity tags, semantic roles, semantically similar words and the likelihood that the sentence makes sense (grammatically and semanti- cally) using a language model. The entire network is trained jointly on all these tasks using weight-sharing, an instance of multitask learning. All the tasks use labeled data ex- cept the language model which is learnt from unlabeled text and represents a novel form of semi-supervised learning for the shared tasks. We show how both multitask learning and semi-supervised learning improve the general- ization of the shared tasks, resulting in state- of-the-art performance.
We address the rating-inference problem, wherein rather than simply decide whether a review is "thumbs up" or "thumbs down", as in previous sentiment analysis work, one must determine an author's evaluation with respect to a multi-point scale (e.g., one to five "stars"). This task represents an interesting twist on standard multi-class text categorization because there are several different degrees of similarity between class labels; for example, "three stars" is intuitively closer to "four stars" than to "one star". We first evaluate human performance at the task. Then, we apply a meta-algorithm, based on a metric labeling formulation of the problem, that alters a given n-ary classifier's output in an explicit attempt to ensure that similar items receive similar labels. We show that the meta-algorithm can provide significant improvements over both multi-class and regression versions of SVMs when we employ a novel similarity measure appropriate to the problem.
A simple but tough-to-beat baseline for sentence embeddings
- Sanjeev Arora
- Yingyu Liang
- Tengyu Ma
Sanjeev Arora, Yingyu Liang, and Tengyu Ma. A simple but tough-to-beat baseline for sentence embeddings. In Proc. of ICLR, 2017.
Raquel Urtasun, and Sanja Fidler. Skip-thought vectors
- Ryan Kiros
- Yukun Zhu
- Ruslan Salakhutdinov
- Richard S Zemel
- Antonio Torralba
Ryan Kiros, Yukun Zhu, Ruslan Salakhutdinov, Richard S. Zemel, Antonio Torralba, Raquel Urtasun, and
Sanja Fidler. Skip-thought vectors. In Proc. of NIPS, pages 3276-3284, 2015.
A sentimental education: Sentiment analysis using subjectivity summarization based on minimum cuts
- Bo Pang
- Lillian Lee
Bo Pang and Lillian Lee. A sentimental education: Sentiment analysis using subjectivity summarization
based on minimum cuts. In Proceedings of the ACL, 2004.