We show how faceted search using a combination of traditional classification systems and mixed-membership topic models can go beyond keyword search to inform resource discovery, hypothesis formulation, and argument extraction for interdisciplinary research. Our test domain is the history and philosophy of scientific work on animal mind and cognition. The methods can be generalized to other research areas and ultimately support a system for semi-automatic identification of argument structures. We provide a case study for the application of the methods to the problem of identifying and extracting arguments about anthropomorphism during a critical period in the development of comparative psychology. We show how a combination of classification systems and mixed-membership models trained over large digital libraries can inform resource discovery in this domain. Through a novel approach of "drill-down" topic modeling---simultaneously reducing both the size of the corpus and the unit of analysis---we are able to reduce a large collection of fulltext volumes to a much smaller set of pages within six focal volumes containing arguments of interest to historians and philosophers of comparative psychology. The volumes identified in this way did not appear among the first ten results of the keyword search in the HathiTrust digital library and the pages bear the kind of "close reading" needed to generate original interpretations that is the heart of scholarly work in the humanities. Zooming back out, we provide a way to place the books onto a map of science originally constructed from very different data and for different purposes. The multilevel approach advances understanding of the intellectual and societal contexts in which writings are interpreted.
Topic models jointly learn topics and document-level topic distribution. Extrinsic evaluation of topic models tends to focus exclusively on topic-level evaluation, e.g. by assessing the coherence of topics. We demonstrate that there can be large discrepancies between topic- and document-level model quality, and that basing model evaluation on topic-level analysis can be highly misleading. We propose a method for automatically predicting topic model quality based on analysis of document-level topic allocations, and provide empirical evidence for its robustness.
Topic models such as Latent Dirichlet Allocation (LDA) have been widely used in information retrieval for tasks ranging from smoothing and feedback methods to tools for exploratory search and discovery. However, classical methods for inferring topic models do not scale up to the massive size of today's publicly available Web-scale data sets. The state-of-the-art approaches rely on custom strategies, implementations and hardware to facilitate their asynchronous, communication-intensive workloads. We present APS-LDA, which integrates state-of-the-art topic modeling with cluster computing frameworks such as Spark using a novel asynchronous parameter server. Advantages of this integration include convenient usage of existing data processing pipelines and eliminating the need for disk writes as data can be kept in memory from start to finish. Our goal is not to outperform highly customized implementations, but to propose a general high-performance topic modeling framework that can easily be used in today's data processing pipelines. We compare APS-LDA to the existing Spark LDA implementations and show that our system can, on a 480-core cluster, process up to 135 times more data and 10 times more topics without sacrificing model quality.
Word embedding models such as Skip-gram learn a vector-space representation for each word, based on the local word collocation patterns that are observed in a text corpus. Latent topic models, on the other hand, take a more global view, looking at the word distributions across the corpus to assign a topic to each word occurrence. These two paradigms are complementary in how they represent the meaning of word occurrences. While some previous works have already looked at using word embeddings for improving the quality of latent topics, and conversely, at using latent topics for improving word embeddings, such "two-step" methods cannot capture the mutual interaction between the two paradigms. In this paper, we propose STE, a framework which can learn word embeddings and latent topics in a unified manner. STE naturally obtains topic-specific word embeddings, and thus addresses the issue of polysemy. At the same time, it also learns the term distributions of the topics, and the topic distributions of the documents. Our experimental results demonstrate that the STE model can indeed generate useful topic-specific word embeddings and coherent latent topics in an effective and efficient way.
Topic models have been successfully applied in lexicon extraction. However, most previous methods are limited to document-aligned data. In this paper, we try to address two challenges of applying topic models to lexicon extraction in non-parallel data: 1) hard to model the word relationship and 2) noisy seed dictionary. To solve these two challenges, we propose two new bilingual topic models to better capture the semantic information of each word while discriminating the multiple translations in a noisy seed dictionary. We extend the scope of topic models by inverting the roles of "word" and "document". In addition, to solve the problem of noise in seed dictionary, we incorporate the probability of translation selection in our models. Moreover, we also propose an effective measure to evaluate the similarity of words in different languages and select the optimal translation pairs. Experimental results using real world data demonstrate the utility and efficacy of the proposed models.
Topic models have been extensively used to organize and interpret the contents of large, unstructured corpora of text documents. Although topic models often perform well on traditional training vs. test set evaluations, it is often the case that the results of a topic model do not align with human interpretation. This interpretability fallacy is largely due to the unsupervised nature of topic models, which prohibits any user guidance on the results of a model. In this paper, we introduce a semi-supervised method called topic supervised non-negative matrix factorization (TS-NMF) that enables the user to provide labeled example documents to promote the discovery of more meaningful semantic structure of a corpus. In this way, the results of TS-NMF better match the intuition and desired labeling of the user. The core of TS-NMF relies on solving a non-convex optimization problem for which we derive an iterative algorithm that is shown to be monotonic and convergent to a local optimum. We demonstrate the practical utility of TS-NMF on the Reuters and PubMed corpora, and find that TS-NMF is especially useful for conceptual or broad topics, where topic key terms are not well understood. Although identifying an optimal latent structure for the data is not a primary objective of the proposed approach, we find that TS-NMF achieves higher weighted Jaccard similarity scores than the contemporary methods, (unsupervised) NMF and latent Dirichlet allocation, at supervision rates as low as 10% to 20%.
Continuous latent time series models are prevalent in Bayesian modeling; examples include the Kalman filter, dynamic collaborative filtering, or dynamic topic models. These models often benefit from structured, non mean field variational approximations that capture correlations between time steps. Black box variational inference with reparameterization gradients (BBVI) allows us to explore a rich new class of Bayesian non-conjugate latent time series models; however, a naive application of BBVI to such structured variational models would scale quadratically in the number of time steps. We describe a BBVI algorithm analogous to the forward-backward algorithm which instead scales linearly in time. It allows us to efficiently sample from the variational distribution and estimate the gradients of the ELBO. Finally, we show results on the recently proposed dynamic word embedding model, which was trained using our method.
The cooperative hierarchical structure is a common and significant data structure observed in, or adopted by, many research areas, such as: text mining (author-paper-word) and multi-label classification (label-instance-feature). Renowned Bayesian approaches for cooperative hierarchical structure modeling are mostly based on topic models. However, these approaches suffer from a serious issue in that the number of hidden topics/factors needs to be fixed in advance and an inappropriate number may lead to overfitting or underfitting. One elegant way to resolve this issue is Bayesian nonparametric learning, but existing work in this area still cannot be applied to cooperative hierarchical structure modeling. In this paper, we propose a cooperative hierarchical Dirichlet process (CHDP) to fill this gap. Each node in a cooperative hierarchical structure is assigned a Dirichlet process to model its weights on the infinite hidden factors/topics. Together with measure inheritance from hierarchical Dirichlet process, two kinds of measure cooperation, i.e., superposition and maximization, are defined to capture the many-to-many relationships in the cooperative hierarchical structure. Furthermore, two constructive representations for CHDP, i.e., stick-breaking and international restaurant process, are designed to facilitate the model inference. Experiments on synthetic and real-world data with cooperative hierarchical structures demonstrate the properties and the ability of CHDP for cooperative hierarchical structure modeling and its potential for practical application scenarios.
Supervisory signals have the potential to make low-dimensional data representations, like those learned by mixture and topic models, more interpretable and useful. We propose a framework for training latent variable models that explicitly balances two goals: recovery of faithful generative explanations of high-dimensional data, and accurate prediction of associated semantic labels. Existing approaches fail to achieve these goals due to an incomplete treatment of a fundamental asymmetry: the intended application is always predicting labels from data, not data from labels. Our prediction-constrained objective for training generative models coherently integrates loss-based supervisory signals while enabling effective semi-supervised learning from partially labeled data. We derive learning algorithms for semi-supervised mixture and topic models using stochastic gradient descent with automatic differentiation. We demonstrate improved prediction quality compared to several previous supervised topic models, achieving predictions competitive with high-dimensional logistic regression on text sentiment analysis and electronic health records tasks while simultaneously learning interpretable topics.
Entity retrieval is the task of finding entities such as people or products in response to a query, based solely on the textual documents they are associated with. Recent semantic entity retrieval algorithms represent queries and experts in finite-dimensional vector spaces, where both are constructed from text sequences. We investigate entity vector spaces and the degree to which they capture structural regularities. Such vector spaces are constructed in an unsupervised manner without explicit information about structural aspects. For concreteness, we address these questions for a specific type of entity: experts in the context of expert finding. We discover how clusterings of experts correspond to committees in organizations, the ability of expert representations to encode the co-author graph, and the degree to which they encode academic rank. We compare latent, continuous representations created using methods based on distributional semantics (LSI), topic models (LDA) and neural networks (word2vec, doc2vec, SERT). Vector spaces created using neural methods, such as doc2vec and SERT, systematically perform better at clustering than LSI, LDA and word2vec. When it comes to encoding entity relations, SERT performs best.
Familia is an open-source toolkit for pragmatic topic modeling in industry. Familia abstracts the utilities of topic modeling in industry as two paradigms: semantic representation and semantic matching. Efficient implementations of the two paradigms are made publicly available for the first time. Furthermore, we provide off-the-shelf topic models trained on large-scale industrial corpora, including Latent Dirichlet Allocation (LDA), SentenceLDA and Topical Word Embedding (TWE). We further describe typical applications which are successfully powered by topic modeling, in order to ease the confusions and difficulties of software engineers during topic model selection and utilization.
In this paper we present the approach of introducing thesaurus knowledge into probabilistic topic models. The main idea of the approach is based on the assumption that the frequencies of semantically related words and phrases, which are met in the same texts, should be enhanced: this action leads to their larger contribution into topics found in these texts. We have conducted experiments with several thesauri and found that for improving topic models, it is useful to utilize domain-specific knowledge. If a general thesaurus, such as WordNet, is used, the thesaurus-based improvement of topic models can be achieved with excluding hyponymy relations in combined topic models.
What defines a visual style? Fashion styles emerge organically from how people assemble outfits of clothing, making them difficult to pin down with a computational model. Low-level visual similarity can be too specific to detect stylistically similar images, while manually crafted style categories can be too abstract to capture subtle style differences. We propose an unsupervised approach to learn a style-coherent representation. Our method leverages probabilistic polylingual topic models based on visual attributes to discover a set of latent style factors. Given a collection of unlabeled fashion images, our approach mines for the latent styles, then summarizes outfits by how they mix those styles. Our approach can organize galleries of outfits by style without requiring any style labels. Experiments on over 100K images demonstrate its promise for retrieving, mixing, and summarizing fashion images by their style.
One of the main computational and scientific challenges in the modern age is to extract useful information from unstructured texts. Topic models are one popular machine-learning approach which infers the latent topical structure of a collection of documents. Despite their success --- in particular of its most widely used variant called Latent Dirichlet Allocation (LDA) --- and numerous applications in sociology, history, and linguistics, topic models are known to suffer from severe conceptual and practical problems, e.g. a lack of justification for the Bayesian priors, discrepancies with statistical properties of real texts, and the inability to properly choose the number of topics. Here we obtain a fresh view on the problem of identifying topical structures by relating it to the problem of finding communities in complex networks. This is achieved by representing text corpora as bipartite networks of documents and words. By adapting existing community-detection methods -- using a stochastic block model (SBM) with non-parametric priors -- we obtain a more versatile and principled framework for topic modeling (e.g., it automatically detects the number of topics and hierarchically clusters both the words and documents). The analysis of artificial and real corpora demonstrates that our SBM approach leads to better topic models than LDA in terms of statistical model selection. More importantly, our work shows how to formally relate methods from community detection and topic modeling, opening the possibility of cross-fertilization between these two fields.
The CL-SciSumm 2016 shared task introduced an interesting problem: given a document D and a piece of text that cites D, how do we identify the text spans of D being referenced by the piece of text? The shared task provided the first annotated dataset for studying this problem. We present an analysis of our continued work in improving our system's performance on this task. We demonstrate how topic models and word embeddings can be used to surpass the previously best performing system.
Embarrassingly (communication-free) parallel Markov chain Monte Carlo (MCMC) methods are commonly used in learning graphical models. However, MCMC cannot be directly applied in learning topic models because of the quasi-ergodicity problem caused by multimodal distribution of topics. In this paper, we develop an embarrassingly parallel MCMC algorithm for sLDA. Our algorithm works by switching the order of sampled topics combination and labeling variable prediction in sLDA, in which it overcomes the quasi-ergodicity problem because high-dimension topics that follow a multimodal distribution are projected into one-dimension document labels that follow a unimodal distribution. Our empirical experiments confirm that the out-of-sample prediction performance using our embarrassingly parallel algorithm is comparable to non-parallel sLDA while the computation time is significantly reduced.
Topic models, and more specifically the class of Latent Dirichlet Allocation (LDA), are widely used for probabilistic modeling of text. MCMC sampling from the posterior distribution is typically performed using a collapsed Gibbs sampler. We propose a parallel sparse partially collapsed Gibbs sampler and compare its speed and efficiency to state-of-the-art samplers for topic models on five well-known text corpora of differing sizes and properties. In particular, we propose and compare two different strategies for sampling the parameter block with latent topic indicators. The experiments show that the increase in statistical inefficiency from only partial collapsing is smaller than commonly assumed, and can be more than compensated by the speedup from parallelization and sparsity on larger corpora. We also prove that the partially collapsed samplers scale well with the size of the corpus. The proposed algorithm is fast, efficient, exact, and can be used in more modeling situations than the ordinary collapsed sampler.
In this paper, we present hierarchical relationbased latent Dirichlet allocation (hrLDA), a data-driven hierarchical topic model for extracting terminological ontologies from a large number of heterogeneous documents. In contrast to traditional topic models, hrLDA relies on noun phrases instead of unigrams, considers syntax and document structures, and enriches topic hierarchies with topic relations. Through a series of experiments, we demonstrate the superiority of hrLDA over existing topic models, especially for building hierarchies. Furthermore, we illustrate the robustness of hrLDA in the settings of noisy data sets, which are likely to occur in many practical scenarios. Our ontology evaluation results show that ontologies extracted from hrLDA are very competitive with the ontologies created by domain experts.
Topic models can provide us with an insight into the underlying latent structure of a large corpus of documents. A range of methods have been proposed in the literature, including probabilistic topic models and techniques based on matrix factorization. However, in both cases, standard implementations rely on stochastic elements in their initialization phase, which can potentially lead to different results being generated on the same corpus when using the same parameter values. This corresponds to the concept of "instability" which has previously been studied in the context of $k$-means clustering. In many applications of topic modeling, this problem of instability is not considered and topic models are treated as being definitive, even though the results may change considerably if the initialization process is altered. In this paper we demonstrate the inherent instability of popular topic modeling approaches, using a number of new measures to assess stability. To address this issue in the context of matrix factorization for topic modeling, we propose the use of ensemble learning strategies. Based on experiments performed on annotated text corpora, we show that a K-Fold ensemble strategy, combining both ensembles and structured initialization, can significantly reduce instability, while simultaneously yielding more accurate topic models.
Besides the text content, documents and their associated words usually come with rich sets of meta informa- tion, such as categories of documents and semantic/syntactic features of words, like those encoded in word embeddings. Incorporating such meta information directly into the generative process of topic models can improve modelling accuracy and topic quality, especially in the case where the word-occurrence information in the training data is insufficient. In this paper, we present a topic model, called MetaLDA, which is able to leverage either document or word meta information, or both of them jointly. With two data argumentation techniques, we can derive an efficient Gibbs sampling algorithm, which benefits from the fully local conjugacy of the model. Moreover, the algorithm is favoured by the sparsity of the meta information. Extensive experiments on several real world datasets demonstrate that our model achieves comparable or improved performance in terms of both perplexity and topic quality, particularly in handling sparse texts. In addition, compared with other models using meta information, our model runs significantly faster.