Topic Features Research Articles

Brand Potential User Identification Algorithm Based on Sentiment Analysis.

This paper firstly compares the current research status of text sentiment analysis and potential customer identification, and introduces the process of building sentiment dictionaries and feature selection, feature screening, and common classification algorithms in text analysis. Secondly, around the most used tool for sentiment analysis, sentiment dictionary, the sentiment polarity discriminative rules of sentiment words are studied. In response to the shortcomings of using a single recognition algorithm in the current process of building sentiment dictionaries, an improved integration rule is designed and an automatic construction method for domain sentiment dictionaries in the social media environment is proposed. Then, this paper analyzes the sentiment topic information existing in user-generated content and adds the domain sentiment lexicon to the joint sentiment topic model as a posteriori information to extract the sentiment topic features, based on which the feature engineering study of potential customer identification is conducted and the feature set is constructed. In addition, a sample resampling method and a diverse integration framework for unbalanced data are designed to work together for the prospect identification task under data skewing in response to the category imbalance in real data. Finally, an experimental study is conducted using a social media text corpus to validate the proposed method in this paper. The proposed domain sentiment lexicon construction method and the joint domain sentiment topic-based lead identification method show good performance in different control group experiments. This paper provides an in-depth study on the construction of domain sentiment lexicon and imbalance classification in theory and provides solutions for companies to discover potential customers in practice, which has certain theoretical significance and practical value.

Popularity prediction for marketer-generated content: A text-guided attention neural network for multi-modal feature fusion

In this paper, we focus on the popularity prediction for marketer-generated content (MGC), which has not been investigated by current studies. To address this problem, we propose a novel deep learning approach, namely the text-guided attention neural network (TGANN) model, to make full use of heterogeneous and multi-modal data related to MGCs (e.g., text descriptions and images). In the proposed model, we first design a filter-based topic model to filter out the noise words and extract topic features from textual descriptions. To lessen the influence of irrelevant information in images, we then propose a text-guided attention mechanism to use the text's topic features to guide the image region representations. Lastly, to determine the contribution of each topic and each image, the TGANN model introduces the attention computation for each visual modality and textual modality. Our experiments are conducted on two real-world datasets of MGC. The quantitative results show that the proposed model outperforms several state-of-the-art methods. The qualitative experiments demonstrate that our model can accurately capture topic attentions, image attentions, and image region attentions. The proposed model provides important practical implications for marketers and online platforms, e.g., estimating the success of online advertising campaigns, creating more attractive marketing contents, and improving recommendation systems.

A Study on Diverse Methods and Performance Measures in Sentiment Analysis

: With the vast development of internet technology 2.0, millions of people are sharing their opinions on different social networking sites. To obtain the necessary information from the huge volume of user-generated data, the attention on sentiment analysis among the research community is growing. Growth and prominence of sentiment analysis is synchronized with an increase in social media and networking sites. Users generally use natural language for speaking, writing, and expressing their views based on various sentiment orientations, ratings, and the features of different products, topics, and issues. This helps to produce ambiguity at the end of the customer's decision based on criticism to form an opinion based on such comments. To overcome the challenges of user-generated content such as noisy, irrelevant information and fake reviews, there is a significant demand for an effective methodology that emphasizes the need for sentiment analysis. This study presents an exhaustive survey of the existing methodologies and highlights the challenges and performance factors of various approaches of sentiment analysis including text preprocessing, opinion spam detection, and aspect level sentiment analysis. Background: User-generated content is growing all over the globe and people more eagerly express their views on social media towards various aspects. The opinionated text is difficult to interpret and arrive at a conclusion based on the feedback gathered from reviews on various sites. Hence, the significance of sentiment analysis is growing to analyze the usergenerated data. Objective: The paper presents an exhaustive review that provides an overview of the pros and cons of the existing techniques and highlights the current techniques in sentiment analysis namely text pre-processing, opinion spam detection, and aspect level sentiment analysis based on machine learning and deep learning. This will be useful to researchers who focus on the challenges very specifically and identify the most common challenges to work forward for a new solution.

Emotion Analysis of Cross-Media Writing Text in the Context of Big Data.

Since the beginning of the 21st century, sentiment analysis has been one of the most active research fields in natural language processing. Now sentiment analysis technology has not only achieved significant results in academia, but also has been widely used in practice. From business services to political campaigns, sentiment analysis is used in more and more fields. Sentiment analysis is essentially to dig out the user’s emotional attitude from the massive emotional natural language text data, and analyze the emotional dynamics of the text author through certain technical means. At present, there is almost no sentiment analysis in cross-media writing content, and it can rarely help cross-media writing vision to advance with the times and comprehensive improvement of writing ability to adapt to the current rapidly developing information society; the commonly used text media in the digital age are not. Then there is the only composition tool. Various new media appearing with the development of the ages continue to intervene in writing, and it is the general trend to cultivate media literacy in writing. The main content of this paper is the research on the emotional intensity of students’ Internet public Aiming at the shortcomings of the topic feature word selection in the sentiment tendency analysis of the students’ Internet public opinion, improvements have been made to facilitate the research on the sentiment intensity of the sentiment analysis of the students’ Internet public opinion. Sentiment analysis of students’ cross-media written text through an improved MapReduce combinator model.

Shapley Idioms: Analysing BERT Sentence Embeddings for General Idiom Token Identification.

This article examines the basis of Natural Language Understanding of transformer based language models, such as BERT. It does this through a case study on idiom token classification. We use idiom token identification as a basis for our analysis because of the variety of information types that have previously been explored in the literature for this task, including: topic, lexical, and syntactic features. This variety of relevant information types means that the task of idiom token identification enables us to explore the forms of linguistic information that a BERT language model captures and encodes in its representations. The core of this article presents three experiments. The first experiment analyzes the effectiveness of BERT sentence embeddings for creating a general idiom token identification model and the results indicate that the BERT sentence embeddings outperform Skip-Thought. In the second and third experiment we use the game theory concept of Shapley Values to rank the usefulness of individual idiomatic expressions for model training and use this ranking to analyse the type of information that the model finds useful. We find that a combination of idiom-intrinsic and topic-based properties contribute to an expression's usefulness in idiom token identification. Overall our results indicate that BERT efficiently encodes a variety of information from topic, through lexical and syntactic information. Based on these results we argue that notwithstanding recent criticisms of language model based semantics, the ability of BERT to efficiently encode a variety of linguistic information types does represent a significant step forward in natural language understanding.

Prediction of Online Psychological Help-Seeking Behavior During the COVID-19 Pandemic: An Interpretable Machine Learning Method.

Online mental health service (OMHS) has been named as the best psychological assistance measure during the COVID-19 pandemic. An interpretable, accurate, and early prediction for the demand of OMHS is crucial to local governments and organizations which need to allocate and make the decision in mental health resources. The present study aimed to investigate the influence of the COVID-19 pandemic on the online psychological help-seeking (OPHS) behavior in the OMHS, then propose a machine learning model to predict and interpret the OPHS number in advance. The data was crawled from two Chinese OMHS platforms. Linguistic inquiry and word count (LIWC), neural embedding-based topic modeling, and time series analysis were utilized to build time series feature sets with lagging one, three, seven, and 14 days. Correlation analysis was used to examine the impact of COVID-19 on OPHS behaviors across different OMHS platforms. Machine learning algorithms and Shapley additive explanation (SHAP) were used to build the prediction. The result showed that the massive growth of OPHS behavior during the COVID-19 pandemic was a common phenomenon. The predictive model based on random forest (RF) and feature sets containing temporal features of the OPHS number, mental health topics, LIWC, and COVID-19 cases achieved the best performance. Temporal features of the OPHS number showed the biggest positive and negative predictive power. The topic features had incremental effects on performance of the prediction across different lag days and were more suitable for OPHS prediction compared to the LIWC features. The interpretable model showed that the increase in the OPHS behaviors was impacted by the cumulative confirmed cases and cumulative deaths, while it was not sensitive in the new confirmed cases or new deaths. The present study was the first to predict the demand for OMHS using machine learning during the COVID-19 pandemic. This study suggests an interpretable machine learning method that can facilitate quick, early, and interpretable prediction of the OPHS behavior and to support the operational decision-making; it also demonstrated the power of utilizing the OMHS platforms as an always-on data source to obtain a high-resolution timeline and real-time prediction of the psychological response of the online public.

Mobile recommender system based on smart city graph

<span>Mobile recommender systems have changed the way people find items, purposes of intrigue, administrations, or even new companions. The innovation behind mobile recommender systems has developed to give client inclinations and social impacts. This paper introduces a first way to build a mobile recommendation system based on smart city graphs that appear topic features, user profiles, and impacts acquired from social connections. It exploits graph centrality measures to expand customized recommendations from the semantic information represented in the graph. The graph shows and chooses graph algorithms for computing chart centrality that is the center of the mobile recommender system are exhibited. Semantic ideas, for example, semantic transcendence and likeness measures, are adjusted to the graph model. Usage challenges confronted to settle execution issues are additionally examined.</span>

An Influence Maximization Algorithm Based on Community-Topic Features for Dynamic Social Networks

Real social networks are huge and continue to expand rapidly. Most existing dynamic influence maximization (IM) algorithms are based on the node-to-node propagation model; hence, they have high time complexity and large storage space consumption. They usually reduce computational complexity using a sampling method while sacrificing the influence spread. In this paper, we propose a topic-aware community independent cascade (IC) model to reduce the complexity of dynamic IM without losing accuracy. The proposed model reduces the problem domain through community-level propagation, and then enhances the global features by integrating community structural features, community topic features, and time information into an IC model. We construct the data structure of the dynamic community index to avoid recalculation when the network grows. Based on the dynamic community index, we design a dynamic IM algorithm to quickly approximate the solution with the (1-1/e)-approximation guarantee. The experimental results on real social networks demonstrated that, compared with existing IM algorithms, the proposed algorithm had better stability and dynamic adaptability, higher computational efficiency, and less space consumption without reducing the approximation ratio and influence spread.

Improving neural topic modeling via Sinkhorn divergence

Textual data have been a major form to convey internet users’ content. How to effectively and efficiently discover latent topics among them has essential theoretical and practical value. Recently, neural topic models(NTMs), especially Variational Auto-encoder-based NTMs, proved to be a successful approach for mining meaningful and interpretable topics. However, they usually suffer from two major issues:(1)Posterior collapse: KL divergence will rapidly reach zeros resulting in low-quality representation in latent distribution; (2)Unconstrained topic generative models: Topic generative models are always unconstrained, which potentially leads to discovering redundant topics. To address these issues, we propose Autoencoding Sinkhorn Topic Model based on Sinkhorn Auto-encoder(SAE) and Sinkhorn divergence. SAE utilizes Sinkhorn divergence rather than problematic KL divergence to optimize the difference between posterior and prior, which is free of posterior collapse. Then, to reduce topic redundancy, Sinkhorn Topic Diversity Regularization(STDR) is presented. STDR leverages the proposed Salient Topic Layer and Sinkhorn divergence for measuring distance between salient topic features and serves as a penalty term in loss function facilitating discovering diversified topics in training. Several experiments have been conducted on 2 popular datasets to verify our contribution. Experiment results demonstrate the effectiveness of the proposed model.

Topical language generation using transformers

AbstractLarge-scale transformer-based language models (LMs) demonstrate impressive capabilities in open-text generation. However, controlling the generated text’s properties such as the topic, style, and sentiment is challenging and often requires significant changes to the model architecture or retraining and fine-tuning the model on new supervised data. This paper presents a novel approach for topical language generation (TLG) by combining a pre-trained LM with topic modeling information. We cast the problem using Bayesian probability formulation with topic probabilities as a prior, LM probabilities as the likelihood, and TLG probability as the posterior. In learning the model, we derive the topic probability distribution from the user-provided document’s natural structure. Furthermore, we extend our model by introducing new parameters and functions to influence the quantity of the topical features presented in the generated text. This feature would allow us to easily control the topical properties of the generated text. Our experimental results demonstrate that our model outperforms the state-of-the-art results on coherency, diversity, and fluency while being faster in decoding.

Integrating Manifold Knowledge for Global Entity Linking with Heterogeneous Graphs

Abstract Entity Linking (EL) aims to automatically link the mentions in unstructured documents to corresponding entities in a knowledge base (KB), which has recently been dominated by global models. Although many global EL methods attempt to model the topical coherence among all linked entities, most of them failed in exploiting the correlations among manifold knowledge helpful for linking, such as the semantics of mentions and their candidates, the neighborhood information of candidate entities in KB and the fine-grained type information of entities. As we will show in the paper, interactions among these types of information are very useful for better characterizing the topic features of entities and more accurately estimating the topical coherence among all the referred entities within the same document. In this paper, we present a novel HEterogeneous Graph-based Entity Linker (HEGEL) for global entity linking, which builds an informative heterogeneous graph for every document to collect various linking clues. Then HEGEL utilizes a novel heterogeneous graph neural network (HGNN) to integrate the different types of manifold information and model the interactions among them. Experiments on the standard benchmark datasets demonstrate that HEGEL can well capture the global coherence and outperforms the prior state-of-the-art EL methods.

Research on the Realization of Travel Recommendations for Different Users Through Deep Learning Under Global Information Management

This article is mainly to study the realization of travel recommendations for different users through deep learning under global information management. The personalized travel route recommendation is realized by establishing personalized travel dynamic interest (PTDR) algorithm and distributed lock manager (DLM) model. It is hoped that this model can provide more complete data information of tourist destinations on the basis of the past, and can also meet the needs of users. The innovation of this article is to compare and analyze with a large number of baseline algorithms, highlighting the superiority of this model in personalized travel recommendation. In addition, the model incorporates the topic factor features, geographic factor features, and user preference features to make the data more in line with user needs and improve the efficiency and applicability of the model. It is hoped that the plan proposed in this article can help users make choices of tourist destinations more conveniently.

Internet Memetics as an Emotiogenic Environment of the Network Communication

The article presents the results of a study that deals with identification of the features peculiar to the functioning of meme communication in the Internet environment in relation to young users on the example of the Russian-language, German-language and English-American-language Internet segments. In the course of this research, the term “memetics” is introduced, which includes various topic structural-systemic and semantic features of the phe-nomenon of “meme” and “meme-communication”. Based on the hermeneutic approach, with the interdisciplinary features of meme-communication in mind, semiotic varieties are analyzed, including both creolized (polycode) and non-creolized (monocode) types of meme-communication inherent in the three above analyzed linguistic segments of the Internet. Topic and compositional features of the obtained data are discussed, taking into account the infor-mation content of the three sources of Internet communication. The main research method is the hermeneutic ap-proach to describing the data obtained, which allows interpreting the results with regard to various departments of knowledge: linguistics, cognitive science, media studies (forms of information transfer), virality, testing, etc. At the center of the study is the concept of “toxicity”, and this makes it possible to determine how communicants-users of the Web can be negatively affected in various ways.

Топос кладбища в романах Е.Г. Водолазкина

The subject of research is the cemetery topos as one of the most important constructs of the artistic space of novels “Laurus”, “Aviator”, “Brisbane”, “Justification of the Island” by the modern Russian writer E.G. Vodolazkin. The relevance of work is explained by the interest of Russian literary studies in the study of literary process of the 20th–21st centuries and the work of one of the brightest, original and authoritative modern prose writers – Eugene Vodolazkin. The purpose of research is to study poetic and functional features of the cemetery topics in the above-mentioned novels. The research methodology is based on a combined approach, including histori-cal and cultural, mythopoetic, structural, comparative and receptive methods. As a result of our research, we establish that Vodolazkin creates a complex, symbolic space that changes and “lives” the novel's story along with the heroes. The cemetery appears in three aspects: as a mythological, commemorative and dialogue space. The image of cemetery is created with the help of smells, light and color, as well as textures (tactile effect), which are also “markers” of positive space or, conversely, space with negative connotations. Thus, the cemetery is one of the main toposes in Vodolazkin’s novels, it is a space for unification, overcoming the “gap” – temporal, spiritual, cultural, linguistic, national, etc. The results obtained in the course of study can be applied in the preparation of university courses on the Russian literature history at the turn of the 20th and 21st centuries, as well as special courses and special seminars.

COVID-19: Detecting depression signals during stay-at-home period.

The new coronavirus outbreak has been officially declared a global pandemic by the World Health Organization. To grapple with the rapid spread of this ongoing pandemic, most countries have banned indoor and outdoor gatherings and ordered their residents to stay home. Given the developing situation with coronavirus, mental health is an important challenge in our society today. In this paper, we discuss the investigation of social media postings to detect signals relevant to depression. To this end, we utilize topic modeling features and a collection of psycholinguistic and mental-well-being attributes to develop statistical models to characterize and facilitate representation of the more subtle aspects of depression. Furthermore, we predict whether signals relevant to depression are likely to grow significantly as time moves forward. Our best classifier yields F-1 scores as high as 0.8 and surpasses the utilized baseline by a considerable margin, 0.173. In closing, we propose several future research avenues.

A New Text Summarization Approach based on Relative Entropy and Document Decomposition

In the era of the fourth industrial revolution, the rapid relay on using the Internet made online resources explosively grow. This revolution emphasized the demand for new approaches to utilize the use of online resources such as texts. Thus, the difficulty to compare unstructured resources (text) is urging the demand of proposing a new approach, which is the core of this paper. In fact, text summarization technology is a vital part of text processing, therefore. The focus is on the semantic information not just on the basic information. It requires mining topic features in order to obtain topic-words and topic-sentences relationships. This automatic text summarization is document decomposition according to relative entropy analysis; which means measuring the difference of the probability distribution to measure the correlation between sentences. This paper introduced a new method for document decomposition, which categorizes the sentences into three types of content. The performance demonstrated the efficiency of using the relative entropy of the topic probability distribution over sentences, which enriched the horizon of text processing and summarization research field.

Unsupervised Graph-Based Tibetan Multi-Document Summarization

Text summarization creates subset that represents the most important or relevant information in the original content, which effectively reduce information redundancy. Recently neural network method has achieved good results in the task of text summarization both in Chinese and English, but the research of text summarization in low-resource languages is still in the exploratory stage, especially in Tibetan. What’s more, there is no large-scale annotated corpus for text summarization. The lack of dataset severely limits the development of low-resource text summarization. In this case, unsupervised learning approaches are more appealing in low-resource languages as they do not require labeled data. In this paper, we propose an unsupervised graph-based Tibetan multi-document summarization method, which divides a large number of Tibetan news documents into topics and extracts the summarization of each topic. Summarization obtained by using traditional graph-based methods have high redundancy and the division of documents topics are not detailed enough. In terms of topic division, we adopt two level clustering methods converting original document into document-level and sentence-level graph, next we take both linguistic and deep representation into account and integrate external corpus into graph to obtain the sentence semantic clustering. Improve the shortcomings of the traditional K-Means clustering method and perform more detailed clustering of documents. Then model sentence clusters into graphs, finally remeasure sentence nodes based on the topic semantic information and the impact of topic features on sentences, higher topic relevance summary is extracted. In order to promote the development of Tibetan text summarization, and to meet the needs of relevant researchers for high-quality Tibetan text summarization datasets, this paper manually constructs a Tibetan summarization dataset and carries out relevant experiments. The experiment results show that our method can effectively improve the quality of summarization and our method is competitive to previous unsupervised methods.

Empower Keywords Generation for Short Texts with Graph-to-Sequence Learning

Keywords are useful in natural language tasks. However, it is a challenge task to extraction keywords from short texts. In which the model may be subject to impaction of topic dependence and poor text organization structure. To resolve this limitation, we propose a keywords generation model ADGCN of short texts based on graph-to-sequence learning. The model to jointly short texts contextual feature and positional feature based adaptation for this task. We learn domain-invariant feature representations by using graph-building feature and node topic feature space, and jointly perform linear generate feature in framework of keywords decoding. Experiment results on real social datasets demonstrate that our proposed model achieves impressive empirical performance on relevance, information and coherence. Besides, the proposed ADGCN also outperforms the state-of-the-arts on public KP20k dataset. The experiments testify that the model can generate the topic keywords of short texts and effectively alleviate the influence of data disturbance.

Danmaku Text Clustering Algorithm Based on Feature Extension and Word-Pair Filtering OBTM

The danmaku text clustering is a hot topic in online video reviews. Given the problem of unsatisfactory clustering accuracy caused by short text and many new words, the danmaku text clustering algorithm based on feature extension and word-pair filtering OBTM is proposed. First, a new-word discovery algorithm based on weight optimization is proposed to retain the features of new words in the danmaku text. Then, the internal information and external knowledge of new words are used to expand the features of the danmaku text for reduced feature sparsity. Furthermore, the OBTM topic model based on word-pair filtering is designed to eliminate noise features. Finally, the Single-Pass algorithm based on cluster center iteration is proposed to obtain the clustering results of topic feature words. Experimental results show that the algorithm proposed in this paper is 13.33 %, 8.52 %, 6.25 % higher than the OBTM, Word2vec+BTM, OurE.Drift* algorithm, respectively, in terms of clustering accuracy.

Некоторые особенности дископатий у собак

Currently, vertebrogenic dyskinesias of various nature are widespread in urban dogs. Among them, there are often herniated discs associated with degenerative processes in various parts of the spinal column in dogs. The article discusses different types of intervertebral disc disease in animals of different breeds and ages, analyzes the frequency of their occurrence, considers the breed characteristics of the distribution of certain pathologies of the spine, topical features of herniated intervertebral discs of various types, as well as risk factors for discopathy in dogs. Among the animals studied, intervertebral disc disease Hansen Type 1 and Hansen Type 2 was more common in dogs of chondrodystrophic breeds (dachshund and French bulldog) and in dogs of large breeds (German shepherd and mestizos). Animals of the older age group are more prone to vertebrogenic dyskinesias. Intervertebral disc injuries were more common in the most mobile parts of the spine, and overweight animals were especially susceptible to them.