Vector Space Model Research Articles

A hunger-based scheduling strategy for distributed crawler

Distributed crawling is one of the mainstream text data collection technologies, which is essential for mining boundless data available on the Internet for users. Internet information is clustered by the correlation of keywords, and users employ search engines to retrieve relevant keywords to get the information they care about. Distributed crawlers are used to mine the Internet information by simulating users’ behavior, the more important keywords that users care about, the higher the correlation of data to keywords. In order to preferentially collect information that users care about with minimal resource consumption, in this paper, we design a scheduling framework and propose a novel scheduling strategy based on hunger for distributed crawler. We first define the load capacity of distributed crawler as hunger which reflects the ability to complete tasks and divide keywords queues into sub-queues based on the hunger of distributed crawlers. Then, we use a vector space model and cosine similarity algorithm to learn the correlation of keywords to text data and apply the optimized logistic algorithm to measure the importance of keywords. Meanwhile, we design a comprehensive evaluation algorithm to quantify the contribution of keywords, so that updating sub-queues order. Finally, new sub-queues are used in the deeper scheduling to preferentially get data that users desire and sacrifice the least number of resources. Experimental results demonstrate that our method optimizes the scheduling procedures and makes crawling more efficient with less run time.

Fuzzy RDF Knowledge Graph Embeddings Through Vector Space Model

Resource description framework (RDF) is a World Wide Web consortium recommendation and has been widely accepted for semantic data models (e.g., knowledge graphs). As such, a large amount of RDF data is readily available. Information in the real world is often uncertain, and uncertain RDF data models have been proposed and received increasing attention. Knowledge graph embedding is an effective way to perform vector projection of structured knowledge, after which we can infer a large-scale knowledge graph. Nowadays, knowledge graph embedding models are widely proposed, but most of them can only handle crisp information. In this article, we concentrate on modeling fuzzy RDF graphs in the vector space. We propose a fuzzy RDF knowledge graph embedding model (FRKGE), which can project the fuzzy entities and fuzzy relations with membership degree to the vector space. Our proposed model preserves the structure of the fuzzy RDF knowledge graph and the explicit expression of membership degree in the vector space, which provides a natural expression of fuzzy RDF tuples. Our experimental results demonstrate that the FRKGE model is competent for the task of representation learning for fuzzy RDF knowledge graphs.

Multimedia information retrieval using artificial neural network

<span lang="EN-US">The importance of the multimedia information retrieval (MIR) is highlighted by the extensive amount of the information on the internet. Image, audio, video, and text are all examples of the characteristics of the raw multimedia data. It is greatly challenging to represent a concept of human perception and how the machine-level language can grasp it (semantic gap of MIR). However, this paper aims to improve the information retrieval model that retrieves data from multimedia. This can be implemented by leveraging the use of variety of algorithms that go through training and testing to extract the model. One of these algorithms extracts text information based on the query language's nature as the vector space model (VSM) and the latent semantic index (LSI) were used. The other technique uses curvelet decomposition and statistic parameters like mean, standard deviation, and signal energy to recover these properties. Additionally, a discrete wavelet transforms (DWT) and signal characteristics-based method is used to retrieve audio signals. Finally, the neural network learning is modeled and trained on a collection of different multimedia images. The learned features have been utilized for presenting a highly sufficient system of multimedia retrieval which operates for a large set of multi-modal datasets. </span>

Study and innovation of effective classification of XML documents using an advanced deep learning approach

<span lang="EN-US">In the digital world, classifying real sensed data in huge volumes derived from numerical problems is a challenging task due to the computational complexity of the metaheuristic searching process. The deep learning approach includes convolutional neural network (CNN), long short-term memory (LSTM), and Bidirectional (BI)-LSTM, suitable for an optimistic processing time of analyzing XML datasets (i.e., social media, trade center, and surveillance data exchanged in the internet world). However, it faces process deviation when datasets extend their range beyond the expected volume. This paper proposes a novel deep learning formwork referred to as archimed improved numerical optimization deep learning (AINODL) to improve the classification of XML datasets. The proposed AINODL framework first extracts feature from XML documents using the vector space model. Secondly, it classifies the XML data using the inbuilt function of the AINODL framework. The experiments demonstrate that the performance parameters accuracy (90%), sensitivity (93%), and specificity (94%) of the proposed AINODL framework are significantly enhanced compared with the existing approaches CNN, LSTM, and BI-LSTM.</span>

Constructing a Shariah Document Screening Prototype Based on Serverless Architecture

The aim of this research is to discuss the groundwork of building an Islamic Banking Document Screening Prototype based on a serverless architecture framework. This research first forms an algorithm for document matching based Vector Space Model (VCM) and adopts Levenshtein Distance for similarity setting. Product proposals will become a query, and policy documents by the central bank will be a corpus or database for document matching. Both the query and corpus went through preprocessing stage prior to similarity analysis. One set of queries with two sets of corpora is tested in this research to compare similarity values. Finally, a prototype of Shariah Document Screening is built based on a serverless architecture framework and ReactJS interface. This research is the first attempt to introduce a Shariah document screening prototype based on a serverless architecture technology that would be useful to the Islamic financial industry towards achieving a Shariah-compliant business. Given the development of Fintech, the output of this research study would be a complement to the existing Fintech applications, which focus on ensuring the Islamic nature of the businesses.

Automated clustering and switching algorithms applied to semantic verbal fluency data in schizophrenia spectrum disorders

ABSTRACT In the cognitive assessment of Schizophrenia Spectrum Disorders (SSD), the standard scoring method for Verbal Fluency (VF) tasks is the number of correct words produced. Finer-grained measures, such as the size of semantic clusters and the number of transitions between them, have been proposed to characterise the cognitive functions involved, but results based on human ratings are heterogeneous. The objective of this study was to develop a computational procedure based on Vector Space Models (VSMs) to assess the predictive ability of these fine-grained measures for class membership in SSD. A semantic VF task was administered to thirty-five people with SSD and a matched group of healthy participants, and their VF productions were characterised manually and using a set of ad-hoc algorithms. Computational estimates consistently showed higher predictive accuracy than models built on VF measures computed by a human rater and models built on the sole total number of words.

An Ensemble and Multi-View Clustering Method Based on Kolmogorov Complexity

The ability to build more robust clustering from many clustering models with different solutions is relevant in scenarios with privacy-preserving constraints, where data features have a different nature or where these features are not available in a single computation unit. Additionally, with the booming number of multi-view data, but also of clustering algorithms capable of producing a wide variety of representations for the same objects, merging clustering partitions to achieve a single clustering result has become a complex problem with numerous applications. To tackle this problem, we propose a clustering fusion algorithm that takes existing clustering partitions acquired from multiple vector space models, sources, or views, and merges them into a single partition. Our merging method relies on an information theory model based on Kolmogorov complexity that was originally proposed for unsupervised multi-view learning. Our proposed algorithm features a stable merging process and shows competitive results over several real and artificial datasets in comparison with other state-of-the-art methods that have similar goals.

An Emotional Analysis Method Based on Multi Model Ensemble Learning

<p>Traditional machine learning models generally use weak supervision model, which is difficult to adapt to the scene of multi classification for emotional text. Therefore, a multi model ensemble learning algorithm for emotional text classification is proposed. The algorithm takes the labeled emotional text data as the training sample, uses the improved TF-IDF algorithm to train the word vector space model, selects three weakly supervised machine learning algorithms, linear SVC, xgboost and logistic regression, to construct the base classifier, and uses the random forest algorithm to construct the meta classifier. It realizes the function of dividing emotional text into three categories: positive, neutral and negative. From the simulation and test results, the AUC values of the multi model ensemble learning algorithm model for each category are 0.93, 0.94 and 1.00, and the AP values are 0.87, 0.86 and 1.00, and the indicators of accuracy and recall are better than the single machine learning model, which realizes the high performance and high accuracy for emotional text classification.</p> <p> </p>

Automatic quantitative assessment of English writing proficiency based on multi-feature fusion

In the existing quantitative evaluation methods of English writing level, the accuracy of feature extraction is low and the error is high. An automatic quantitative evaluation method of English writing level based on multi-feature fusion is put forward. By using vector space model and Jekard similarity coefficient to determine the cosine similarity of English text, the features of English text are extracted by Manhattan distance. Through kernel function, the multi-feature fusion of English writing text is realised. The multivariate linear regression model is used to determine the feature weight of English text and to quantitatively process the feature data. The automatic quantitative evaluation model of English writing level is constructed to complete the automatic quantitative evaluation of English writing level. The experimental results show that the accuracy of the proposed automatic quantitative evaluation method is always higher than 90 and the minimum error of text feature extraction is about 2%.

Improved Electro Search Algorithm with Intelligent Controller Control System: ESPID Algorithm

Studies have established that hybrid models outperform single models. The particle swarm algorithm (PSO)-based PID (proportional-integral-derivative) controller control system is used in this study to determine the parameters that directly impact the speed and performance of the Electro Search (ESO) algorithm to obtain the global optimum point. ESPID algorithm was created by integrating this system with the ESO algorithm. The improved ESPID algorithm has been applied to 7 multi-modal benchmark test functions. The acquired results were compared to those derived using the ESO, PSO, Atom Search Optimization (ASO), and Vector Space Model (VSM) algorithms. As a consequence, it was determined that the ESPID algorithm’s mean score was superior in all functions. Additionally, while comparing the mean duration value and standard deviations, it is observed that it is faster than the ESO algorithm and produces more accurate results than other algorithms. ESPID algorithm has been used for the least cost problem in the production of pressure vessels, which is one of the real-life problems. Statistical results were compared with ESO, Genetic algorithm and ASO. ESPID was found to be superior to other methods with the least production cost value of 5885.452.

A Topic-Specific Web Crawler using Deep Convolutional Networks

This paper presented a new focused crawler that efficiently supports the Turkish language. The developed architecture was divided into multiple units: a control unit, crawler unit, link extractor unit, link sorter unit, and natural language processing unit. The crawler's units can work in parallel to process the massive amount of published websites. Also, the proposed Convolutional Neural Network (CNN) based natural language processing unit can professionally classifying Turkish text and web pages. Extensive experiments using three datasets have been performed to illustrate the performance of the developed approach. The first dataset contains 50,000 Turkish web pages downloaded by the developed crawler, while the other two are publicly available and consist of “28,567” and “22,431” Turkish web pages, respectively. In addition, the Vector Space Model (VSM) in general and word embedding state-of-the-art techniques, in particular, were investigated to find the most suitable one for the Turkish language. Overall, results indicated that the developed approach had achieved good performance, robustness, and stability when processing the Turkish language. Also, Bidirectional Encoder Representations from Transformer (BERT) was found to be the most appropriate embedding for building an efficient Turkish language classification system. Finally, our experiments showed superior performance of the developed natural language processing unit against seven state-of-the-art CNN classification systems. Where accuracy improvement compared to the second-best is 10% and 47% compared to the lowest performance.

A Novel Patent Knowledge Extraction Method for Innovative Design

As an important source of inspiration, the great number of patent documents provides designers with valuable knowledge of design rationale (DR), including issues, intent, pros and cons of the solutions. Researchers have carried out a number of data analysis studies based on patent information, which is now a new discipline called Patinformatics, including the analysis of patent information from a macro perspective and the identification and extraction of patent knowledge from a micro perspective. If DR knowledge could be extracted automatically from the patent documents and provided to designers as a source of inspiration, it would greatly promote innovative design, and at the same time promote the reuse of patent documents and the wide application of DR theory, which can be like killing three birds with one stone. To address this issue, this study proposes an improved lexical-syntactic pattern method for DR centric patent knowledge extraction, including DR Vector Space model (DRVS), DRV Trigger Word (DRV-TW), Design Rationale Vector (DRV), DR credibility (DRC) and others, and DRV based knowledge extraction algorithms. Knowledge extraction experiments were conducted on 1491 patent documents to verify the feasibility and performance of the method. In addition, two other sets of comparative experiments were conducted using the FastText and BERT machine learning methods, and the results further confirmed the reliability of the proposed method for low-resource corpus.

Retracted: Clustering Merchants and Accurate Marketing of Products Using the Segmentation Tree Vector Space Model

Retracted: Clustering Merchants and Accurate Marketing of Products Using the Segmentation Tree Vector Space Model

Difference Analysis in Risk Focus Based on Text Disclosure Frequency: A Triple Perspectives of Bank Managers, Financial Analysts, and Credit Raters

This study aims to explore the differences in risk focus among bank managers, financial analysts, and credit raters by comparing the similarity of text frequency of bank annual reports, financial analysts' reports, and credit rating reports. We use the Vector Space model to vectorize the text. And we conducted the Cosine similarity measurements to get the variation of annual risk focus change of unilateral perspective and risk focus difference of the three parties. In the empirical analysis, we collect 204 Form 10-K filings, 193 credit rating reports and 193 credit rating reports from 2006 to 2022. The empirical results show that the bank risk focus perceived by bank managers, financial analysts and credit raters changes over time. Besides, the similarities of risk focus among bank managers, financial analysts and credit raters these three parties are quite low, especially for financial analysts and credit raters. Thus, our empirical findings found that bank managers, financial analysts and credit raters have different judgements on risk. It indicates that the comprehensive use of the information disclosed by the three parties can provide a more comprehensive evaluation of bank risk.

Word Embeddings and Semantic Spaces in Natural Language Processing

One of the critical hurdles, and breakthroughs, in the field of Natural Language Processing (NLP) in the last two decades has been the development of techniques for text representation that solves the so-called curse of dimensionality, a problem which plagues NLP in general given that the feature set for learning starts as a function of the size of the language in question, upwards of hundreds of thousands of terms typically. As such, much of the research and development in NLP in the last two decades has been in finding and optimizing solutions to this problem, to feature selection in NLP effectively. This paper looks at the development of these various techniques, leveraging a variety of statistical methods which rest on linguistic theories that were advanced in the middle of the last century, namely the distributional hypothesis which suggests that words that are found in similar contexts generally have similar meanings. In this survey paper we look at the development of some of the most popular of these techniques from a mathematical as well as data structure perspective, from Latent Semantic Analysis to Vector Space Models to their more modern variants which are typically referred to as word embeddings. In this review of algoriths such as Word2Vec, GloVe, ELMo and BERT, we explore the idea of semantic spaces more generally beyond applicability to NLP.

Research On Text Classification Based On Deep Neural Network

Text classification is one of the classic tasks in the field of natural language processing. The goal is to identify the category to which the text belongs. Text categorization is widely used in email detection, sentiment analysis, topic marking and other fields. However, good text representation is the key to improve the performance of natural language processing tasks such as text classification. Traditional text representation adopts bag-of-words model or vector space model, which not only loses the context information of the text, but also faces the problems of high latitude and high sparsity. In recent years, with the increase of data and the improvement of computing performance, the use of deep learning technology to represent and classify texts has attracted great attention. Convolutional neural network, recurrent neural network and recurrent neural network with attention mechanism are used to represent the text, and then to classify the text and other natural language processing tasks, all of which have better performance than the traditional methods. In this paper, we design two sentence-level text representation and classification models based on the deep network. The details are as follows: (1) Text representation and classification model based on bidirectional cyclic and convolutional neural networks-BRCNN. Brcnn's input is the word vector corresponding to each word in the sentence; After using cyclic neural network to extract word order information in sentences, convolution neural network is used to extract higher-level features of sentences. After convolution, the maximum pool operation is used to obtain sentence vectors. At last, softmax classifier is used for classification. Cyclic neural network can capture the word order information in sentences, while convolutional neural network can extract useful features. Experiments on eight text classification tasks show that BRCNN model can get better text feature representation, and the classification accuracy rate is equal to or higher than that of the prior art.. (2) A text representation and classification model based on attention mechanism and convolutional neural network-ACNN. ACNN model uses the recurrent neural network with attention mechanism to obtain the context vector; Then convolution neural network is used to extract more advanced feature information. The maximum pool operation is adopted to obtain a sentence vector; At last, the softmax classifier is used to classify the text. Experiments on eight text classification benchmark data sets show that ACNN improves the stability of model convergence, and can converge to an optimal or local optimal solution better than BRCNN.

Topic Modelling Using VSM-LDA For Document Summarization

Summarization is a process to simplify the contents of a document by eliminating elements that are considered unimportant but do not reduce the core meaning the document wants to convey. However, as is known, a document will contain more than one topic. So it is necessary to identify the topic so that the summarization process is more effective. Latent Dirichlet Allocation (LDA) is a commonly used method of identifying topics. However, when running a program on a different dataset, LDA experiences "order effects", that is, the resulting topic will be different if the train data sequence is changed. In the same document input, LDA will provide inconsistent topics resulting in low coherence values. Therefore, this paper proposes a topic modelling method using a combination of LDA and VSM (Vector Space Model) for automatic summarization. The proposed method can overcome order effects and identify document topics that are calculated based on the TF-IDF weight on VSM generated by LDA. The results of the proposed topic modeling method on the 1300 Twitter data resulted in the highest coherence value reaching 0.72. The summary results obtained Rouge 1 is 0.78, Rouge 2 is 0.67 dan Rouge L is 0.80.

Semantic-Based K-Means Clustering for IMDB Top 100 Movies

Textual documents are growing rapidly through the internet in today’s modern technology era. Electronic structured databases archive offline and online documents, e-mails, webpages, blog and social network posts. Without appropriate ranking and demand clustering when there is classification without any specifics, it is quite difficult to retain and access these documents. K-means is one of the methods that is frequently used for clustering. In terms of determining the proximity of meaning or semantics between data, the distance-based K-means method still has flaws. To get around this issue, semantic similarity can be estimated by measuring the level of similarity between objects in a cluster. This research provides a method for clustering documents based on semantic similarity. The approach is carried out by defining document synopses from the IMDB and Wikipedia databases using the NLTK dictionary, and we provide a semantic-based K-means clustering approach that assesses not only the similarity of the data represented as a vector space model with TFIDF, but also the semantic similarity of the data Precision, recall, and F-measure, we demonstrate how well the semantic-based K-means clustering technique works using experimental findings from the IMDB and Wikipedia top 100 movies datasets.

Mengiplementasikan Vector Space Model Similarity Euclidean Distance Menggunakan TFIDF Pada Klasifikasi Teks Bahasa Indonesia

Weighting based on the term with stemming techniques to get the basic word form term in question. This will the application of the Indonesian language text classification machine using the K-Nearest Neighbor algorithm and the Vector Space Model method on the TFIDF frequency weighting of the number of words and the Euclidean Distance function. comparison between the test documents and the test sample collection Using news documents as learning documents, a total of 10 (10) documents with 3 (three) categories, produces an Precision and Recall 90.00% for k = 5 using frequency weighting in words with the Euclidean Distance function.

Identification and Filtering of Web Spams Using a Machine Learning Method

In order to enhance the filtering of spam on the Internet and improve the experience of Internet users, this paper proposed to convert the email text into vector features using the vector space model, constructed a two-dimensional matrix, and used a convolutional neural network (CNN) to identify spam on the Internet. The CNN was compared with other two classifiers, support vector machine (SVM), and backward-propagation neural network (BPNN), in simulation experiments. The final results showed that the spam recognition algorithm with CNN as the classifier had better recognition performance than the algorithms with SVM and BPNN classifiers and was also more advantageous in terms of recognition cost and time for spam; in addition, the CNN had the best recognition performance when the number of extracted features was 15.