Hierarchical topic modeling is a potentially powerful instrument for determining the topical structure of text collections that allows constructing a topical hierarchy representing levels of topical abstraction. However, tuning of parameters of hierarchical models, including the number of topics on each hierarchical level, remains a challenging task and an open issue. In this paper, we propose a Renyi entropy-based approach for a partial solution to the above problem. First, we propose a Renyi entropy-based metric of quality for hierarchical models. Second, we propose a practical concept of hierarchical topic model tuning tested on datasets with human mark-up. In the numerical experiments, we consider three different hierarchical models, namely, hierarchical latent Dirichlet allocation (hLDA) model, hierarchical Pachinko allocation model (hPAM), and hierarchical additive regularization of topic models (hARTM). We demonstrate that hLDA model possesses a significant level of instability and, moreover, the derived numbers of topics are far away from the true numbers for labeled datasets. For hPAM model, the Renyi entropy approach allows us to determine only one level of the data structure. For hARTM model, the proposed approach allows us to estimate the number of topics for two hierarchical levels.
Electronic Health Record (EHR) data can be represented as discrete counts over a high dimensional set of possible procedures, diagnoses, and medications. Supervised topic models present an attractive option for incorporating EHR data as features into a prediction problem: given a patient's record, we estimate a set of latent factors that are predictive of the response variable. However, existing methods for supervised topic modeling struggle to balance prediction quality and coherence of the latent factors. We introduce a novel approach, the prediction-focused topic model, that uses the supervisory signal to retain only features that improve, or do not hinder, prediction performance. By removing features with irrelevant signal, the topic model is able to learn task-relevant, interpretable topics. We demonstrate on a EHR dataset and a movie review dataset that compared to existing approaches, prediction-focused topic models are able to learn much more coherent topics while maintaining competitive predictions.
Although response generation (RG) diversification for single-turn dialogs has been well developed, it is less investigated for natural multi-turn conversations. Besides, past work focused on diversifying responses without considering topic coherence to the context, producing uninformative replies. In this paper, we propose the Topic-coherent Hierarchical Recurrent Encoder-Decoder model (THRED) to diversify the generated responses without deviating the contextual topics for multi-turn conversations. In overall, we build a sequence-to-sequence net (Seq2Seq) to model multi-turn conversations. And then we resort to the latent Variable Hierarchical Recurrent Encoder-Decoder model (VHRED) to learn global contextual distribution of dialogs. Besides, we construct a dense topic matrix which implies word-level correlations of the conversation corpora. The topic matrix is used to learn local topic distribution of the contextual utterances. By incorporating both the global contextual distribution and the local topic distribution, THRED produces both diversified and topic-coherent replies. In addition, we propose an explicit metric (\emph{TopicDiv}) to measure the topic divergence between the post and generated response, and we also propose an overall metric combining the diversification metric (\emph{Distinct}) and \emph{TopicDiv}. We evaluate our model comparing with three baselines (Seq2Seq, HRED and VHRED) on two real-world corpora, respectively, and demonstrate its outstanding performance in both diversification and topic coherence.
We report on the potential for using algorithms for non-negative matrix factorization (NMF) to improve parameter estimation in topic models. While several papers have studied connections between NMF and topic models, none have suggested leveraging these connections to develop new algorithms for fitting topic models. Importantly, NMF avoids the "sum-to-one" constraints on the topic model parameters, resulting in an optimization problem with simpler structure and more efficient computations. Building on recent advances in optimization algorithms for NMF, we show that first solving the NMF problem then recovering the topic model fit can produce remarkably better fits, and in less time, than standard algorithms for topic models. While we focus primarily on maximum likelihood estimation, we show that this approach also has the potential to improve variational inference for topic models. Our methods are implemented in the R package fastTopics.
Overcomplete latent representations have been very popular for unsupervised feature learning in recent years. In this paper, we specify which overcomplete models can be identified given observable moments of a certain order. We consider probabilistic admixture or topic models in the overcomplete regime, where the number of latent topics can greatly exceed the size of the observed word vocabulary. While general overcomplete topic models are not identifiable, we establish generic identifiability under a constraint, referred to as topic persistence. Our sufficient conditions for identifiability involve a novel set of "higher order" expansion conditions on the topic-word matrix or the population structure of the model. This set of higher-order expansion conditions allow for overcomplete models, and require the existence of a perfect matching from latent topics to higher order observed words. We establish that random structured topic models are identifiable w.h.p. in the overcomplete regime. Our identifiability results allows for general (non-degenerate) distributions for modeling the topic proportions, and thus, we can handle arbitrarily correlated topics in our framework. Our identifiability results imply uniqueness of a class of tensor decompositions with structured sparsity which is contained in the class of Tucker decompositions, but is more general than the Candecomp/Parafac (CP) decomposition.
In this paper we introduce the problem of determining the topic that a set of images is describing, where every topic is represented as a set of words. Different from other problems like tag assignment or similar, a) we assume multiple images are used as input instead of single image, b) Input images are typically not visually related, c) Input images are not necessarily semantically close, and d) Output word space is unconstrained. In our proposed solution, visual information of each query image is used to retrieve similar images with text labels (tags) from an image database. We consider a scenario where the tags are very noisy and diverse, given that they were obtained by implicit crowd-sourcing in a database of 1 million images and over seventy seven thousand tags. The words or tags associated to each query are processed jointly in a word selection algorithm using random walks that allows to refine the search topic, rejecting words that are not part of the topic and produce a set of words that fairly describe the topic. Experiments on a dataset of 300 topics, with up to twenty images per topic, show that our algorithm performs better than the proposed baseline for any number of query images. We also present a new Conditional Random Field (CRF) word mapping algorithm that preserves the semantic similarity of the mapped words, increasing the performance of the results over the baseline.
We introduce jet topics: a framework to identify underlying classes of jets from collider data. Because of a close mathematical relationship between distributions of observables in jets and emergent themes in sets of documents, we can apply recent techniques in "topic modeling" to extract jet topics from data with minimal or no input from simulation or theory. As a proof of concept with parton shower samples, we apply jet topics to determine separate quark and gluon jet distributions for constituent multiplicity. We also determine separate quark and gluon rapidity spectra from a mixed Z-plus-jet sample. While jet topics are defined directly from hadron-level multi-differential cross sections, one can also predict jet topics from first-principles theoretical calculations, with potential implications for how to define quark and gluon jets beyond leading-logarithmic accuracy. These investigations suggest that jet topics will be useful for extracting underlying jet distributions and fractions in a wide range of contexts at the Large Hadron Collider.
Topic drift is a common phenomenon in multi-turn dialogue. Therefore, an ideal dialogue generation models should be able to capture the topic information of each context, detect the relevant context, and produce appropriate responses accordingly. However, existing models usually use word or sentence level similarities to detect the relevant contexts, which fail to well capture the topical level relevance. In this paper, we propose a new model, named STAR-BTM, to tackle this problem. Firstly, the Biterm Topic Model is pre-trained on the whole training dataset. Then, the topic level attention weights are computed based on the topic representation of each context. Finally, the attention weights and the topic distribution are utilized in the decoding process to generate the corresponding responses. Experimental results on both Chinese customer services data and English Ubuntu dialogue data show that STAR-BTM significantly outperforms several state-of-the-art methods, in terms of both metric-based and human evaluations.
We propose the split-diffuse (SD) algorithm that takes the output of an existing dimension reduction algorithm, and distributes the data points uniformly across the visualization space. The result, called the topic grids, is a set of grids on various topics which are generated from the free-form text content of any domain of interest. The topic grids efficiently utilizes the visualization space to provide visual summaries for massive data. Topical analysis, comparison and interaction can be performed on the topic grids in a more perceivable way.
Topic models have been widely explored as probabilistic generative models of documents. Traditional inference methods have sought closed-form derivations for updating the models, however as the expressiveness of these models grows, so does the difficulty of performing fast and accurate inference over their parameters. This paper presents alternative neural approaches to topic modelling by providing parameterisable distributions over topics which permit training by backpropagation in the framework of neural variational inference. In addition, with the help of a stick-breaking construction, we propose a recurrent network that is able to discover a notionally unbounded number of topics, analogous to Bayesian non-parametric topic models. Experimental results on the MXM Song Lyrics, 20NewsGroups and Reuters News datasets demonstrate the effectiveness and efficiency of these neural topic models.
Add to Chrome
Add to Firefox