Random-walk term weighting for improved text classification

This paper describes a new approach for estimating term weights in a document, and shows how the new weighting scheme can be used to improve the accuracy of a text classifier. The method uses term co-occurrence as a measure of dependency between word features. A random-walk model is applied on a graph encoding words and co-occurrence dependencies, resulting in scores that represent a quantification of how a particular word feature contributes to a given context. Experiments performed on three standard classification datasets show that the new random-walk based approach outperforms the traditional term frequency approach of feature weighting.

TFIDF algorithm was used for feature weighting in text classification. But the result of classification was not very well because of lack of class information in feature weighting. The known class information in the training set was used to improve the traditional TFIDF feature weight algorithm. Class distinction ability and class description ability were introduced, respectively expressed by inverse class frequency and term frequency in class, document frequency in class. A new feature weight algorithm based on class information, TF_IDT, was proposed. Naive Bayes classifier was used to test the algorithm. The precision, recall and F1 measure were significantly increased. Macro F1 measure raise by 6.46%. It was proved to be useful for improving text classification to use class information in feature weighting. In addition, the computational complexity of the proposed algorithm was lower and more suitable for use in fields of limited computing capability. Keywords-text classification; feature weight; inverse class frequency; term frequency in class; document frequency in class

TFIDF algorithm was used for feature weighting in text classification. But the result of classification was not very well because of lack of class information in feature weighting. The known class information in the training set was used to improve the traditional TFIDF feature weight algorithm. Class distinction ability and class description ability were introduced, respectively expressed by inverse class frequency and term frequency in class, document frequency in class. A new feature weight algorithm based on class information, TF_IDT, was proposed. Naïve Bayes classifier was used to test the algorithm. The precision, recall and F1 measure were significantly increased. Macro F1 measure raise by 6.46%. It was proved to be useful for improving text classification to use class information in feature weighting. In addition, the computational complexity of the proposed algorithm was lower and more suitable for use in fields of limited computing capability.

Even if recent Transformer-based architectures, such as BERT, achieved impressive results in semantic processing tasks, their fine-tuning stage still requires large scale training resources. Usually, Data Augmentation (DA) techniques can help to deal with low resource settings. In Text Classification tasks, the objective of DA is the generation of well-formed sentences that i) represent the desired task category and ii) are novel with respect to existing sentences. In this paper, we propose a neural approach to automatically learn to generate new examples using a pre-trained sequence-to-sequence model. We first learn a task-oriented similarity function that we use to pair similar examples. Then, we use these example pairs to train a model to generate examples. Experiments in low resource settings show that augmenting the training material with the proposed strategy systematically improves the results on text classification and natural language inference tasks by up to 10% accuracy, outperforming existing DA approaches.

Knowledge enhanced pre-trained language models (K-PLMs) are shown to be effective for many public tasks in the literature, but few of them have been successfully applied in practice. To address this problem, we propose K-AID, a systematic approach that includes a low-cost knowledge acquisition process for acquiring domain knowledge, an effective knowledge infusion module for improving model performance, and a knowledge distillation component for reducing the model size and deploying K-PLMs on resource-restricted devices (e.g., CPU) for real-world application. Importantly, instead of capturing entity knowledge like the majority of existing K-PLMs, our approach captures relational knowledge, which contributes to better improving sentence-level text classification and text matching tasks that play a key role in question answering (QA). We conducted a set of experiments on five text classification tasks and three text matching tasks from three domains, namely E-commerce, Government, and Film&TV, and performed online A/B tests in E-commerce. Experimental results show that our approach is able to achieve substantial improvement on sentence-level question answering tasks and bring beneficial business value in industrial settings.

How to use advanced deep learning technology to build an effective and powerful text classification model, extract text semantic attributes and achieve good classification results on large-scale test datasets has always been the main problem and severe challenge of text classification tasks. The purpose of this paper is to study the application optimization of NLP system in text semantics and text classification based on deep learning technology. A WC-GCN classification algorithm combining GCN and Attention mechanism is proposed. The related tasks of text classification are discussed, including three typical text classification tasks, including sentiment analysis, topic analysis, and news classification. The algorithm model proposed in this paper is verified from the experimental point of view. Compared with other deep learning-based text classification methods, the proposed algorithm model performs better on the public datasets of three text classification tasks. The news text also achieved a high accuracy rate.

Many applications require semantic understanding of short texts, and inferring discriminative and coherent latent topics is a critical and fundamental task in these applications. Conventional topic models largely rely on word co-occurrences to derive topics from a collection of documents. However, due to the length of each document, short texts are much more sparse in terms of word co-occurrences. Recent studies show that the Dirichlet Multinomial Mixture (DMM) model is effective for topic inference over short texts by assuming that each piece of short text is generated by a single topic. However, DMM has two main limitations. First, even though it seems reasonable to assume that each short text has only one topic because of its shortness, the definition of “shortness” is subjective and the length of the short texts is dataset dependent. That is, the single-topic assumption may be too strong for some datasets. To address this limitation, we propose to model the topic number as a Poisson distribution, allowing each short text to be associated with a small number of topics (e.g., one to three topics). This model is named PDMM. Second, DMM (and also PDMM) does not have access to background knowledge (e.g., semantic relations between words) when modeling short texts. When a human being interprets a piece of short text, the understanding is not solely based on its content words, but also their semantic relations. Recent advances in word embeddings offer effective learning of word semantic relations from a large corpus. Such auxiliary word embeddings enable us to address the second limitation. To this end, we propose to promote the semantically related words under the same topic during the sampling process, by using the generalized Pólya urn (GPU) model. Through the GPU model, background knowledge about word semantic relations learned from millions of external documents can be easily exploited to improve topic modeling for short texts. By directly extending the PDMM model with the GPU model, we propose two more effective topic models for short texts, named GPU-DMM and GPU-PDMM. Through extensive experiments on two real-world short text collections in two languages, we demonstrate that PDMM achieves better topic representations than state-of-the-art models, measured by topic coherence. The learned topic representation leads to better accuracy in a text classification task, as an indirect evaluation. Both GPU-DMM and GPU-PDMM further improve topic coherence and text classification accuracy. GPU-PDMM outperforms GPU-DMM at the price of higher computational costs.

Short text classification plays an important role in NLP and its applications span a wide range of activities such as sentiment analysis, spam detection. Recently, attention mechanism is widely used in text classification task. Inspired by this, a text classification model based on multi-attention network(MAN) is proposed in this study, which perform well in extracting information related to text category. In our model, we combine the textual information based on multi-attention mechanism, which enables model to focus on global information of the sentence. We tested effectiveness of our model using several standard text classification datasets. Experiment told that our model achieved state-of-the-art results on all datasets.

Text classification is one of the fundamental tasks in natural language processing. Recently, deep neural networks have achieved promising performance in the text classification task compared to shallow models. Despite of the significance of deep models, they ignore the fine-grained (matching signals between words and classes) classification clues since their classifications mainly rely on the text-level representations. To address this problem, we introduce the interaction mechanism to incorporate word-level matching signals into the text classification task. In particular, we design a novel framework, EXplicit interAction Model (dubbed as EXAM), equipped with the interaction mechanism. We justified the proposed approach on several benchmark datasets including both multilabel and multi-class text classification tasks. Extensive experimental results demonstrate the superiority of the proposed method. As a byproduct, we have released the codes and parameter settings to facilitate other researches.

Despite being a well‐known problem, feature weighting and feature selection are a major predicament for clustering. Most of the algorithms, which provide weighting or selection of features, require the number of clusters to be known in advance. On the other hand, the existing automatic clustering procedures that can determine the number of clusters are computationally expensive and often do not make a room for feature weighting or selection. In this paper, we propose a Gibbs sampling‐based algorithm for the Dirichlet process mixture model, which can determine the number of clusters and can also incorporate a near‐optimal feature weighting. We show that in the limiting case, the algorithm approaches a hard clustering procedure, which resembles minimization of an underlying clustering objective similar to weighted k‐means with an additional forfeit for the number of clusters and hence retains the simplicity of the Llyod's heuristics. To avoid the trivial solution of the resulting linear program, we include an additional entropic penalty on the feature weights. The proposed algorithm is tested on several synthetic and real‐life datasets. Through a detailed experimental analysis, we demonstrate the competitiveness of our proposal against the baseline as well as state‐of‐the‐art procedures for centre‐based high‐dimensional clustering.

Feature weighting is considered as an important machine learning approach to deal with the problem of estimating the quality of attributes for pattern classification applications. Local Hyperlinear Learning based Relief (LH-Relief) was shown to be very efficient in estimating attributions in high-dimensional data involving irrelevant noises. However, the convergence of LH-Relief can not be guaranteed. In this paper, we propose an innovative feature weighting algorithm to solve the problem of LH-Relief, called Iterative Local Hyperlinear Learning based Relief (ILH-Relief). ILH-Relief is based on LH-Relief using classical margin maximization. The key idea is to estimate the feature weights through local approximation and gradient descent. To demonstrate the viability and the effectiveness of our formulation for feature selection in supervised learning, we perform extensive experiments on both UCI and Microarray datasets. The proposed algorithm can save at least half of feature ranking time with a better classification performance compared with other feature weighting methods.

Image retrieval is the process of finding images in the database that are similar to the query image by measuring how close the feature values of the query image are to other images. Image retrieval is currently dominated by approaches that combine several different representations or features. The optimal weight of each feature is needed in combining the image features such as color features, texture features, and shape features. In this study, we use a multi-layer perceptron artificial neural network (MLP) method to obtain feature weights automatically and simultaneously look for optimal weights. The color moment is used to find nine color features, Gray Level Co-occurrence Matrix (GLCM) to find four texture features, and Hu Moment to find seven shape features totaling 20 neurons and all of these features become the input layer in our MLP network. Three neurons in output layers become the automatic weight of each feature. These weights are used to combine each feature's role in obtaining the relevant image. Euclidean Distance is used to measure similarity. The average precision values obtained using automatic feature weights are 93.94% for the synthetic dataset, 91.19% for the Coil-100 dataset, and 54.31% for the Wang dataset. These results have an average difference of 5.06% with the target so automatic feature weighting works well. This value is obtained at a hidden layer size of 11 and a learning rate of 0.1. In addition, the use of automatic feature weighting gives more accurate results compared to manual feature weighting.

Text classification is a fundamental task in many applications such as topic labeling, sentiment analysis, and spam detection. The text syntactic relationship and word sequence are important and useful for text classification. How to model and incorporate them to improve performance is one key challenge. Inspired by human behavior in understanding text. In this paper, we combine the syntactic relationship, sequence structure, and semantics for text representation, and propose an attention-enhanced capsule network-based text classification model. Specifically, we use graph convolutional neural networks to encode syntactic dependency trees, build multi-head attention to encode dependencies relationship in text sequence, merge with semantic information by capsule network at last. Extensive experiments on five datasets demonstrate that our approach can effectively improve the performance of text classification compared with state-of-the-art methods. The result also shows capsule network, graph convolutional neural network, and multi-headed attention has integration effects on text classification tasks.

For many text classification tasks, there is a major problem posed by the lack of labeled data in a target domain. Although classifiers for a target domain can be trained on labeled text data from a related source domain, the accuracy of such classifiers is usually lower in the cross-domain setting. Recently, string kernels have obtained state-of-the-art results in various text classification tasks such as native language identification or automatic essay scoring. Moreover, classifiers based on string kernels have been found to be robust to the distribution gap between different domains. In this paper, we formally describe an algorithm composed of two simple yet effective transductive learning approaches to further improve the results of string kernels in cross-domain settings. By adapting string kernels to the test set without using the ground-truth test labels, we report significantly better accuracy rates in cross-domain English polarity classification.

Recent research on model interpretability in natural language processing extensively uses feature scoring methods for identifying which parts of the input are the most important for a model to make a prediction (i.e. explanation or rationale). However, previous research has shown that there is no clear best scoring method across various text classification tasks while practitioners typically have to make several other ad-hoc choices regarding the length and the type of the rationale (e.g. short or long, contiguous or not). Inspired by this, we propose a simple yet effective and flexible method that allows selecting optimally for each data instance: (1) a feature scoring method; (2) the length; and (3) the type of the rationale. Our method is inspired by input erasure approaches to interpretability which assume that the most faithful rationale for a prediction should be the one with the highest difference between the model's output distribution using the full text and the text after removing the rationale as input respectively. Evaluation on four standard text classification datasets shows that our proposed method provides more faithful, comprehensive and highly sufficient explanations compared to using a fixed feature scoring method, rationale length and type. More importantly, we demonstrate that a practitioner is not required to make any ad-hoc choices in order to extract faithful rationales using our approach.