skip to main content

Title: BATS: A Spectral Biclustering Approach to Single Document Topic Modeling and Segmentation
Existing topic modeling and text segmentation methodologies generally require large datasets for training, limiting their capabilities when only small collections of text are available. In this work, we reexamine the inter-related problems of “topic identification” and “text segmentation” for sparse document learning, when there is a single new text of interest. In developing a methodology to handle single documents, we face two major challenges. First is sparse information : with access to only one document, we cannot train traditional topic models or deep learning algorithms. Second is significant noise : a considerable portion of words in any single document will produce only noise and not help discern topics or segments. To tackle these issues, we design an unsupervised, computationally efficient methodology called Biclustering Approach to Topic modeling and Segmentation (BATS). BATS leverages three key ideas to simultaneously identify topics and segment text: (i) a new mechanism that uses word order information to reduce sample complexity, (ii) a statistically sound graph-based biclustering technique that identifies latent structures of words and sentences, and (iii) a collection of effective heuristics that remove noise words and award important words to further improve performance. Experiments on six datasets show that our approach outperforms several state-of-the-art more » baselines when considering topic coherence, topic diversity, segmentation, and runtime comparison metrics. « less
; ; ; ; ; ;
Award ID(s):
1942680 1952085
Publication Date:
Journal Name:
ACM Transactions on Intelligent Systems and Technology
Page Range or eLocation-ID:
1 to 29
Sponsoring Org:
National Science Foundation
More Like this
  1. Ruis, Andrew ; Lee, Seung B. (Ed.)
    When text datasets are very large, manually coding line by line becomes impractical. As a result, researchers sometimes try to use machine learning algorithms to automatically code text data. One of the most popular algorithms is topic modeling. For a given text dataset, a topic model provides probability distributions of words for a set of “topics” in the data, which researchers then use to interpret meaning of the topics. A topic model also gives each document in the dataset a score for each topic, which can be used as a non-binary coding for what proportion of a topic is in the document. Unfortunately, it is often difficult to interpret what the topics mean in a defensible way, or to validate document topic proportion scores as meaningful codes. In this study, we examine how keywords from codes developed by human experts were distributed in topics generated from topic modeling. The results show that (1) top keywords of a single topic often contain words from multiple human-generated codes; and conversely, (2) words from human-generated codes appear as high-probability keywords in multiple topic. These results explain why directly using topics from topic models as codes is problematic. However, they also imply that topicmore »modeling makes it possible for researchers to discover codes from short word lists.« less
  2. Classification and topic modeling are popular techniques in machine learning that extract information from large-scale datasets. By incorporating a priori information such as labels or important features, methods have been developed to perform classification and topic modeling tasks; however, most methods that can perform both do not allow for guidance of the topics or features. In this paper, we propose a novel method, namely Guided Semi-Supervised Non-negative Matrix Factorization (GSSNMF), that performs both classification and topic modeling by incorporating supervision from both pre-assigned document class labels and user-designed seed words. We test the performance of this method on legal documents provided by the California Innocence Project and the 20 Newsgroups dataset. Our results show that the proposed method improves both classification accuracy and topic coherence in comparison to past methods such as Semi-Supervised Non-negative Matrix Factorization (SSNMF), Guided Non-negative Matrix Factorization (Guided NMF), and Topic Supervised NMF.
  3. Dalalyan, Aynak (Ed.)
    Topic models have become popular tools for dimension reduction and exploratory analysis of text data which consists in observed frequencies of a vocabulary of p words in n documents, stored in a p×n matrix. The main premise is that the mean of this data matrix can be factorized into a product of two non-negative matrices: a p×K word-topic matrix A and a K×n topic-document matrix W. This paper studies the estimation of A that is possibly element-wise sparse, and the number of topics K is unknown. In this under-explored context, we derive a new minimax lower bound for the estimation of such A and propose a new computationally efficient algorithm for its recovery. We derive a finite sample upper bound for our estimator, and show that it matches the minimax lower bound in many scenarios. Our estimate adapts to the unknown sparsity of A and our analysis is valid for any finite n, p, K and document lengths. Empirical results on both synthetic data and semi-synthetic data show that our proposed estimator is a strong competitor of the existing state-of-the-art algorithms for both non-sparse A and sparse A, and has superior performance is many scenarios of interest.
  4. Networked data involve complex information from multifaceted channels, including topology structures, node content, and/or node labels etc., where structure and content are often correlated but are not always consistent. A typical scenario is the citation relationships in scholarly publications where a paper is cited by others not because they have the same content, but because they share one or multiple subject matters. To date, while many network embedding methods exist to take the node content into consideration, they all consider node content as simple flat word/attribute set and nodes sharing connections are assumed to have dependency with respect to all words or attributes. In this paper, we argue that considering topic-level semantic interactions between nodes is crucial to learn discriminative node embedding vectors. In order to model pairwise topic relevance between linked text nodes, we propose topical network embedding, where interactions between nodes are built on the shared latent topics. Accordingly, we propose a unified optimization framework to simultaneously learn topic and node representations from the network text contents and structures, respectively. Meanwhile, the structure modeling takes the learned topic representations as conditional context under the principle that two nodes can infer each other contingent on the shared latent topics.more »Experiments on three real-world datasets demonstrate that our approach can learn significantly better network representations, i.e., 4.1% improvement over the state-of-the-art methods in terms of Micro-F1 on Cora dataset. (The source code of the proposed method is available through the github link: https://« less
  5. null (Ed.)
    Topic modeling, a method for extracting the underlying themes from a collection of documents, is an increasingly important component of the design of intelligent systems enabling the sense-making of highly dynamic and diverse streams of text data related but not limited to scientific discovery. Traditional methods such as Dynamic Topic Modeling (DTM) do not lend themselves well to direct parallelization because of dependencies from one time step to another. In this paper, we introduce and empirically analyze Clustered Latent Dirichlet Allocation (CLDA), a method for extracting dynamic latent topics from a collection of documents. Our approach is based on data decomposition in which the data is partitioned into segments, followed by topic modeling on the individual segments. The resulting local models are then combined into a global solution using clustering. The decomposition and resulting parallelization leads to very fast runtime even on very large datasets. Our approach furthermore provides insight into how the composition of topics changes over time and can also be applied using other data partitioning strategies over any discrete features of the data, such as geographic features or classes of users. In this paper CLDA is applied successfully to seventeen years of NIPS conference papers (2,484 documentsmore »and 3,280,697 words), seventeen years of computer science journal abstracts (533,588 documents and 46,446,184 words), and to forty years of the PubMed corpus (4,025,976 documents and 386,847,695 words). On the PubMed corpus, we demonstrate the versatility of CLDA by segmenting the data by both time and by journal. Our runtime on this corpus demonstrates an ability to function on very large scale datasets.« less