Word2vec Model Research Articles

Exif2Vec: A Framework to Ascertain Untrustworthy Crowdsourced Images Using Metadata

In the context of social media, the integrity of images is often dubious. To tackle this challenge, we introduce Exif2Vec , a novel framework specifically designed to discover modifications in social media images. The proposed framework leverages an image’s metadata to discover changes in an image. We use a service-oriented approach that considers discovery of changes in images as a service . A novel word-embedding-based approach is proposed to discover semantic inconsistencies in an image metadata that are reflective of the changes in an image. These inconsistencies are used to measure the severity of changes. The novelty of the approach resides in that it does not require the use of images to determine the underlying changes. We use a pretrained Word2Vec model to conduct experiments. The model is validated on two different fact-checked image datasets, i.e., images related to general context and a context-specific image dataset. Notably, our findings showcase the remarkable efficacy of our approach, yielding results of up to 80% accuracy. This underscores the potential of our framework.

The counting principle makes number words unique

Abstract Following Ariel (2021. Why it’s hard to construct ad hoc number concepts. In Caterina Mauri, Ilaria Fiorentini, & Eugenio Goria (eds.), Building categories in interaction: Linguistic resources at work, 439–462. Amsterdam: John Benjamins), we argue that number words manifest distinct distributional patterns from open-class lexical items. When modified, open-class words typically take selectors (as in kinda table), which select a subset of their potential denotations (e.g., “nonprototypical table”). They are typically not modified by loosening operators (e.g., approximately), since even if bare, typical lexemes can broaden their interpretation (e.g., table referring to a rock used as a table). Number words, on the other hand, have a single, precise meaning and denotation and cannot take a selector, which would need to select a subset of their (single) denotation (??kinda seven). However, they are often overtly broadened (approximately seven), creating a range of values around N. First, we extend Ariel’s empirical examination to the larger COCA and to Hebrew (HeTenTen). Second, we propose that open-class and number words belong to sparse versus dense lexical domains, respectively, because the former exhibit prototypicality effects, but the latter do not. Third, we further support the contrast between sparse and dense domains by reference to: synchronic word2vec models of sparse and dense lexemes, which testify to their differential distributions, numeral use in noncounting communities, and different renewal rates for the two lexical types.

From word clouds to Word Rain: Revisiting the classic word cloud to visualize climate change texts

Word Rain is a development of the classic word cloud. It addresses some of the limitations of word clouds, in particular the lack of a semantically motivated positioning of the words, and the use of font size as a sole indicator of word prominence. Word Rain uses the semantic information encoded in a distributional semantics-based language model – reduced into one dimension – to position the words along the x-axis. Thereby, the horizontal positioning of the words reflects semantic similarity. Font size is still used to signal word prominence, but this signal is supplemented with a bar chart, as well as with the position of the words on the y-axis. We exemplify the use of Word Rain by three concrete visualization tasks, applied on different real-world texts and document collections on climate change. In these case studies, word2vec models, reduced to one dimension with t-SNE, are used to encode semantic similarity, and TF-IDF is used for measuring word prominence. We evaluate the technique further by carrying out domain expert reviews.

PreSubLncR: Predicting Subcellular Localization of Long Non-Coding RNA Based on Multi-Scale Attention Convolutional Network and Bidirectional Long Short-Term Memory Network

The subcellular localization of long non-coding RNA (lncRNA) provides important insights and opportunities for an in-depth understanding of cell biology, revealing disease mechanisms, drug development, and innovation in the biomedical field. Although several computational methods have been proposed to identify the subcellular localization of lncRNA, it is difficult to accurately predict the subcellular localization of lncRNA effectively with these methods. In this study, a new deep-learning predictor called PreSubLncR has been proposed for accurately predicting the subcellular localization of lncRNA. This predictor firstly used the word embedding model word2vec to encode the RNA sequences, and then combined multi-scale one-dimensional convolutional neural networks with attention and bidirectional long short-term memory networks to capture the different characteristics of various RNA sequences. This study used multiple RNA subcellular localization datasets for experimental validation, and the results showed that our method has higher accuracy and robustness compared with other state-of-the-art methods. It is expected to provide more in-depth insights into cell function research.

Word2Vec and SVM Fusion for Advanced Sentiment Analysis on Amazon Reviews

As a matter of fact, sentiment analysis plays a crucial role in different kinds of reviews for online shopping. With this in mind, this study explores the application of Word2Vec and Support Vector Machine (SVM) in sentiment analysis of Amazon reviews. To be specific, initially, this study conducted preprocessing and feature extraction on a large-scale review dataset, revealing deep semantic relationships between words through the Word2Vec model. Subsequently, this study utilized SVM for model training and optimization, achieving efficient and accurate sentiment classification. According to the analysis, preliminary experimental results indicate that this combined method can effectively capture complex patterns and relationships in the text, demonstrating significant advantages in enhancing the accuracy as well as efficiency of sentiment analysis compared to traditional methods. Overall, these results shed light on providing a powerful tool for e-commerce platforms to understand better and analyze consumer opinions and feelings as well as guiding further exploration.

Emotion detection using Word2Vec and convolution neural networks

<div>Emotion detection from text plays a very critical role in different domains, including customer service, social media analysis, healthcare, financial services, education, human-to-computer interaction, psychology, and many more. Nowadays, deep learning techniques become popular due to their capabilities to capture inherent complex insights and patterns from raw data. In this paper, we have used the Word2Vec embedding approach that takes care of the semantic and contextual understanding of text making it more realistic while detecting emotions. These embeddings act as input to the convolution neural network (CNN) to capture insights using feature maps. The Word2Vec and CNN models applied to the international survey on emotion antecedents and reactions (ISEAR) dataset outperform the models in the literature in terms of accuracy and F1-score as model evaluation metrics. The proposed approach not only obtains high accuracy in emotion detection tasks but also generates interpretable representations that contribute to the understanding of emotions in textual data. These findings carry significant implications for applications in diverse domains, such as social media analysis, market research, clinical assessment and counseling, and tailored recommendation systems.</div>

Validation of a Thai artificial chatmate designed for cheering up the elderly during the COVID-19 pandemic.

The COVID-19 pandemic severely affected populations of all age groups. The elderly are a high-risk group and are highly vulnerable to COVID-19. Assistive software chatbots can enhance the mental health status of the elderly by providing support and companionship. The objective of this study was to validate a Thai artificial chatmate for the elderly during the COVID-19 pandemic and floods. Chatbot design includes the establishment of a dataset and emotional word vectors in which data consisting of emotional sentences were converted into the word vector form using a pre-trained word2vec model. A word vector was then input into a convolutional neural network (CNN) and trained until the model converges using sentence embedding and similarity word segmentation. Sentence vectors were generated by averaging each word vector using an averaged vector method. For approximate similarity matching, the Annoy library was used to create the indices in tree sorting. Data were collected from 22 elderly and assessed by the Post-Study System Usability Questionnaire (PSSUQ). The study revealed that 72.73% of the respondents found the chatbot easy to learn and use, 63.64% of the respondents found the chatbot can autonomously determine the next course of action, and 59.09% of the respondents believed that troubleshooting guidelines were provided for overcoming errors. The accuracy of the chatbot providing a reasonable response is 56.20±13.99%. Most users were satisfied with the chatbot system. The proposed chatbot provided considerable essential insights into the development of assistance systems for the elderly during the coronavirus pandemic (COVID-19) and during the period of national disasters. The model can be expanded to other applications in the future.

Cyberbullying Text Identification based on Deep Learning and Transformer-based Language Models

In the contemporary digital age, social media platforms like Facebook, Twitter, and YouTube serve as vital channels for individuals to express ideas and connect with others. Despite fostering increased connectivity, these platforms have inadvertently given rise to negative behaviors, particularly cyberbullying. While extensive research has been conducted on high-resource languages such as English, there is a notable scarcity of resources for low-resource languages like Bengali, Arabic, Tamil, etc., particularly in terms of language modeling. This study addresses this gap by developing a cyberbullying text identification system called BullyFilterNeT tailored for social media texts, considering Bengali as a test case. The intelligent BullyFilterNeT system devised overcomes Out-of-Vocabulary (OOV) challenges associated with non-contextual embeddings and addresses the limitations of context-aware feature representations. To facilitate a comprehensive understanding, three non-contextual embedding models GloVe, FastText, and Word2Vec are developed for feature extraction in Bengali. These embedding models are utilized in the classification models, employing three statistical models (SVM, SGD, Libsvm), and four deep learning models (CNN, VDCNN, LSTM, GRU). Additionally, the study employs six transformer-based language models: mBERT, bELECTRA, IndicBERT, XML-RoBERTa, DistilBERT, and BanglaBERT, respectively to overcome the limitations of earlier models. Remarkably, BanglaBERT-based BullyFilterNeT achieves the highest accuracy of 88.04% in our test set, underscoring its effectiveness in cyberbullying text identification in the Bengali language.

An MIU-based deep embedded clustering model for urban functional zoning from remote sensing images and VGI data

Urban functional zoning offers valuable insights into urban morphology and sustainable development. However, the conventional fixed spatial units, such as blocks and grids, cannot easily capture the morphological characteristics inherent in functional union and separation during urban evolution. In this paper, by taking advantage of remote sensing images and geospatial big data, we propose a minimum identification unit (MIU)-based urban functional zoning model. This approach integrates the deep embedded clustering of buildings to generate the spatial unit segmentation, and then identifies the urban function by generating semantic vectors with the Word2Vec model. The effectiveness of the proposed method was tested in the city of Wuhan in China. The results highlight that MIUs provide a more flexible and suitable unit for segmenting urban functional zones compared to traditional street blocks. The proposed method is a feasible way to deal with the semantic redundancy of volunteered geographic information (VGI) data when identifying urban function, and the quality issue only has a significant impact on the minor functional types. Moreover, the building clustering results can effectively reveal the fine-scale urban structure, especially for the administration, manufacturing, and residential types. This demonstrates the potential of our approach in enhancing the understanding of urban morphology and supporting sustainable urban development.

Evaluation of word embedding models used for diachronic semantic change analysis

In the last decade, the quantitative analysis of diachronic changes in language and lexical semantic changes have become the subject of active research. A significant role was played by the development of new effective techniques of word embedding. This direction has been effectively demonstrated in a number of studies. Some of them have focused on the analysis of the optimal type of word2vec models, hyperparameters for training, and evaluation techniques. In this research, we used Corpus of Historical American English (COHA). The paper demonstrates the results of multiple training runs and the comparison of word2vec models with different variations of hyperparameters used for lexical semantic change detection. In addition to traditional word similarities and analogical reasoning tests, we used testing on an extended set of synonyms. We have evaluated word2vec models on the set of more than 100,000 English synsets that were randomly selected from the WordNet database. We have shown that changing the word2vec model parameters (such as a dimension of word embedding, a size of context window, a type of model, a word discard rate etc.) can significantly impact on the resulting word embedding vector space and the detected lexical semantic changes. Additionally, the results strongly depended on properties of the corpus, such as word frequency distribution.

Utilizing Data Mining for the Synthesis of Functionalized Tungsten Oxide with Enhanced Oxygen Vacancies for Highly Sensitive Detection of Triethylamine.

The optimal combination of metal ions and ligands for sensing materials was estimated by using a data-driven model developed in this research. This model utilized advanced computational algorithms and a data set of 100,000 literature pieces. The semiconductor metal oxide (SMO) that is most suitable for detecting triethylamine (TEA) with the highest probability was identified by using the Word2vec model, which employed the maximum likelihood method. The loss function of the probability distribution was minimized in this process. Based on the analysis, a novel hierarchical nanostructured tungsten-based coordination with 2,5-dihydroxyterephthalic acid (W-DHTA) was synthesized. This synthesis involved a postsynthetic hydrothermal treatment (psHT) and the self-assembly of tungsten oxide nanorods. The tungsten oxide nanorods had a significant number of oxygen vacancies. Various techniques were used to characterize the synthesized material, and its sensing performance toward volatile organic compound (VOC) gases was evaluated. The results showed that the functionalized tungsten oxide exhibited an exceptionally high sensitivity and selectivity toward TEA gas. Even in a highly disturbed environment, the detection limit for TEA gas was as low as 40 parts per billion (ppb). Furthermore, our findings suggest that the control of oxygen vacancies in sensing materials plays a crucial role in enhancing the sensitivity and selectivity of gas sensors. This approach was supported by the utilization of density functional theory (DFT) computation and machine learning algorithms to assess and analyze the performance of sensor devices, providing a highly efficient and universally applicable research methodology for the development and design of next-generation functional materials.

Validation of a Thai artificial chatmate designed for cheering up the elderly during the COVID-19 pandemic

Background The COVID-19 pandemic severely affected populations of all age groups. The elderly are a high-risk group and are highly vulnerable to COVID-19. Assistive software chatbots can enhance the mental health status of the elderly by providing support and companionship. The objective of this study was to validate a Thai artificial chatmate for the elderly during the COVID-19 pandemic and floods. Methods Chatbot design includes the establishment of a dataset and emotional word vectors in which data consisting of emotional sentences were converted into the word vector form using a pre-trained word2vec model. A word vector was then input into a convolutional neural network (CNN) and trained until the model converges using sentence embedding and similarity word segmentation. Sentence vectors were generated by averaging each word vector using an averaged vector method. For approximate similarity matching, the Annoy library was used to create the indices in tree sorting. Data were collected from 22 elderly and assessed by the Post-Study System Usability Questionnaire (PSSUQ). Results The study revealed that 72.73% of the respondents found the chatbot easy to learn and use, 63.64% of the respondents found the chatbot can autonomously determine the next course of action, and 59.09% of the respondents believed that troubleshooting guidelines were provided for overcoming errors. The accuracy of the chatbot providing a reasonable response is 56.20±13.99%. Conclusions Most users were satisfied with the chatbot system. The proposed chatbot provided considerable essential insights into the development of assistance systems for the elderly during the coronavirus pandemic (COVID-19) and during the period of national disasters. The model can be expanded to other applications in the future.

Differences in Urban Development in China from the Perspective of Point of Interest Spatial Co-Occurrence Patterns

An imbalance in urban development in China has become a contradiction. Points of Interest (POIs) serve as representations of the spatial distribution of urban functions. Analyzing POI spatial co-occurrence patterns can reveal the agglomeration patterns of urban functions across cities at different levels, providing insights into imbalances in urban development. Using POI data from 297 cities in China, the Word2vec model was employed to model the POI spatial co-occurrence patterns, allowing for the quantification of fine-granular urban functionality. Subsequently, the cities were clustered into five tiers representing different levels of development. An urban hierarchical disparity index and graph were introduced to examine variations in urban functions across different tiers. A significant correlation between POI spatial co-occurrence patterns and the GDP of cities at different levels was demonstrated. This study revealed a notable polarization trend characterized by the development of top-tier cities and lagging tail-end cities. Top-tier cities exhibit advantages in terms of their commercial environments, such as international banks, companies, and transportation facilities. Conversely, tail-end cities face deficiencies in urban infrastructure. It is crucial to coordinate resource allocation and establish sustainable development strategies that foster mutual support between the top-tier and tail-end cities.

Technological competitiveness of China's internet platformers: comparison of Google and Baidu by using patent text information

Purpose This study aims to assess the technological capability of Chinese internet platforms (BAT: Baidu, Alibaba, Tencent) compared to US ones (GAFA: Google, Amazon, Facebook, Apple). More specifically, this study explores Baidu’s technological catching-up process with Google by analyzing their patent textual information. Design/methodology/approach The authors retrieved 26,383 Google patents and 6,695 Baidu patents from PATSTAT 2019 Spring version. The collected patent documents were vectorized using the Word2Vec model first, and then K-means clustering was applied to visualize the technological space of two firms. Finally, novel indicators were proposed to capture the technological catching-up process between Baidu and Google. Findings The results show that Baidu follows a trend of US rather than Chinese technology which suggests Baidu is aggressively seeking to catch up with US players in the process of its technological development. At the same time, the impact index of Baidu patents increases over time, reflecting its upgrading of technological competitiveness. Originality/value This study proposed a new method to analyze technology mapping and evolution based on patent text information. As both US and China are crucial players in the internet industry, it is vital for policymakers in third countries to understand the technological capacity and competitiveness of both countries to develop strategic partnerships effectively.

A new word embedding model integrated with medical knowledge for deep learning-based sentiment classification

The development of intelligent systems that use social media data for decision-making processes in numerous domains such as politics, business, marketing, and finance, has been made possible by the popularity of social media platforms. However, the utilization of textual data from social media in the healthcare management industry is still somewhat limited when it is compared to other industries. Investigating how current machine learning and natural language processing technologies can be used in the healthcare industry to gauge public sentiment is an important study. Earlier works on healthcare sentiment analysis have utilized traditional word embedding models trained on the general and medical corpus. However, integration of medical knowledge to pre-trained word embedding models has not been considered yet. Word embedding models trained on the general corpus led to the problem of lacking medical knowledge and the models trained on the small size of the medical corpus have limitations in capturing semantic and syntactic properties. This research proposes a new word embedding model named Word Embedding Integrated with Medical Knowledge Vector (WE-iMKVec). The proposed model integrates sentiment lexicons and medical knowledgebases into the pre-trained word embedding to enrich the properties of word embedding. A new medical-aware sentiment polarity score is proposed for the utilization in learning neural-network sentiment and these vectors incorporate with the original pre-trained word vectors. The resulting vectors are enriched with lexicon vectors and the medical knowledge vectors: Adverse Drug Reaction (ADR) vector and Unified Medical Language System (UMLS) vector are used to build the proposed WE-iMKVec model. WE-iMKVec is validated on the five different social media healthcare review datasets and the empirical results showed its superiority over traditional word embedding models in medical sentiment analysis. The highest improvement can be found in the patients.info medical condition dataset where the proposed model outperforms three conventional word2vec models (Google-News, PubMed-PMC, and Drug Reviews) by 12.7 %, 31.4 %, and 25.4 % respectively in terms of F1 score.

Тональність художньої прози як вияв особливостей перекладу

The paper is devoted to the peculiarities of sentiment analysis of texts. The necessity to take into account the tone as an important component of the text’s content to achieve equivalence in literary translation and ensure the preservation of the emotional and expressive background and expressiveness of the original is substantiated. The importance of automating the processes of detecting the text’s tone to obtain better and more accurate results is determined. The paper outlines the main methods of preprocessing textual information and identifies a set of the most popular machine learning approaches that can be used for tone classification. In particular, the attention is focused on the Bag-of-Words, TF-IDF Vectorizer, and Word2Vec models used to convert textual data into numerical data. Classifier algorithms are analysed as a further step in data processing (in particular, Bayesian, maximum entropy, support vectors, etc.). The advantages of the LIWC-22 software, which has proven itself and has been successfully tested with the involvement of experts in psychology, sociology and linguistics, are described separately. Using the above-mentioned lexicon, a sentiment analysis of three versions of Y. Vynnychuk's short story («The Embroidered World») translations into English (author's translation by M. Naidan and automated translations by Google and DeepL) is carried out. The total number of words is established, and the indicators of authenticity and tone (in the binary coordinate system «positive/negative») of each translation option are determined. Prospects for further research in the field of sentiment analysis for the interpretation of original and translated texts are formulated.

Enhanced Generation Algorithm for Medical Retrieval Based on Word2Vec Model and RAG Framework

Enhanced Generation Algorithm for Medical Retrieval Based on Word2Vec Model and RAG Framework

A Textual Logical Analysis of the Evolution of Marx’s Urban Critical Thought

Abstract Focusing on elucidating Marx’s urban critical thought is of positive significance for understanding China’s urbanization path and shaping the socialist urban development model with Chinese characteristics in the new era. The main point of Marx’s urban critical thought is the starting point of this paper, and it examines the manifestation and evolution logic of Marx’s urban critical thought. For the textual logic of Marx’s urban critical thought, the Word2vec model is utilized to extract textual features. Then, the BERT model is introduced for text word vector construction, and a multiclassification model of Marx’s urban critical thought text is established based on the fusion of multiple semantic information features. After completing the construction of the method, relevant texts about Marx’s urban critical thought were collected to construct a dataset, and the classification effect was analyzed to explore the evolutionary development trend of Marx’s urban critical thought. The results show that the text classification model can achieve the optimal classification accuracy of 93.15% by applying the 15th layer. The optimal performance is obtained when the local text word length is 100, and the sustainability growth of Marx’s urban critical thought between 1950 and 2020 reaches 102.34%. An in-depth analysis of Marx’s urban critical thought evolutionary logic can provide theoretical guidance for the modernization of Chinese cities.

Research on the Path of Teaching Reform of Chemistry Major in Agricultural and Forestry Colleges and Universities by Integrating Curriculum Civics in the Network Information Age

Abstract In the information age, the ideological and political education of college students has become an important task of higher education. Based on the Word2vec model and the improved TF-IDF algorithm, this paper constructs a text analysis method to analyze the distribution of ideological and political elements in chemistry textbooks taught in colleges and universities. Taking the text of chemistry teaching materials of an agricultural and forestry university as an example, the constructed model is used to mine the distribution of the ideological and political elements and the integration of the ideological and political elements in the chemistry teaching materials of colleges and universities is counted. On this basis, a teaching reform path for integrating Civics and Politics in chemistry courses is proposed, and it is being applied to the freshman class of chemistry majors at an agricultural and forestry university. The effectiveness of integrating Civics into the chemistry program was assessed by comparing the chemistry scores and Civics literacy levels of sample students before and after the teaching practice. 74.3% of the students had a high degree of acceptance of the integration of civics into the chemistry curriculum. The chemistry grades and the level of Civic and Political Literacy of the students in the classes with teaching reform were significantly improved, and the overall scores were 4.26 points and 5.7% higher than those of the control classes respectively. The pedagogical reform path of the curriculum Civics integrated into Chemistry has a significant effect on students’ civics literacy level.

Text-Based Measure of ESG Risk Exposure

Against the backdrop of increasing instability and uncertainty in the world economy, evaluating corporate ESG (Environmental, Social, and Governance) risks has become an important tool for enhancing resilience, effectively countering and mitigating external risks, and achieving sustainable development. In order to assess the magnitude of ESG risk disclosure at the corporate level, this paper first constructs dictionaries for the three dimensions of ESG using Word2vec model and a corpus consisting of ESG reports and ESG-related news. Secondly, a firm-level ESG risk disclosure measure is developed using quarterly earnings conference calls from 2017 to 2024. Subsequently, the validity of the measurement indicators is verified through a review of human reading of phrase-level surface validity and excerpt snippets. Finally, the patterns of corporate ESG risk disclosure are demonstrated from both industry and time perspectives. These results indicate that text-based measurements provide a reliable quantification of corporate-level ESG risk exposure. By measuring corporate ESG risks, potential risks in environmental, social, and governance aspects can be timely identified, thus contributing to proactive risk management.