Latent Semantic Analysis Research Articles

Marathi Extractive Text Summarization using Latent Semantic Analysis and Fuzzy Algorithms

Extractive text summarization involves the retention of only the most important sentences in a document. In the past, multiple approaches involving both statistical and machine learning-based methods have been used for this task. The crucial step in extractive text summarization is getting the right ranking order of sentences in the document in terms of their importance. Singular value decomposition or SVD algorithm based on latent semantic analysis focuses on recognizing the sections in the document which are related in terms of their semantic nature. Fuzzy algorithms involve reasoning of the priority order of the sentences using fuzzy logic unlike the use of discrete values. While significant work has been done for extractive text summarization in English and other foreign languages, there is ample scope for improving the performance of systems when dealing with Marathi text. In this paper, SVD and fuzzy algorithms are proposed for performing extractive text summarization on Marathi documents. Work is done upon the modeling principle, data flow, and parameters of these algorithms such that they are best suited for the task. An analysis of the characteristics of both these techniques is conducted to compare their benefits and shortcomings. The performance of both the algorithms is evaluated on a document dataset using standard performance metrics including the ROUGE metric. An unbiased comparison of both these techniques is carried out to inform the applicability of them, especially when working with Marathi or in general, non-English text.

Extracting marketing information from product reviews: a comparative study of latent semantic analysis and probabilistic latent semantic analysis

Extracting marketing information from product reviews: a comparative study of latent semantic analysis and probabilistic latent semantic analysis

Human Intelligence and Value of Machine Advancements in Cognitive Science A Design thinking Approach

Latent Semantic Analysis (LSA) is an approach used for expressing and extracting textual meanings using statistical evaluations or modeling applied to vast corpora of text, and its development has been a major motivation for this study to understand the design thinking approach. We introduced LSA and gave some instances of how it might be used to further our knowledge of cognition and to develop practical technology. Since LSA's inception, other alternative statistical models for meaning detection and analysis in text corpora have been created, tested, and refined. This study demonstrates the value that statistical models of semantics provide to the study of cognitive science and the development of cognition. These models are particularly useful because they enable researchers to study a wide range of problems pertaining to knowledge, discourse perception, text cognition, and language using expansive representations of human intelligence.

Leveraging Natural Language Processing for Automated Information Inquiry from Building Information Models

Building Information Modeling (BIM) is a trending technology in the building industry that can increase efficiency throughout construction. Various practical information can be obtained from BIM models during the project life cycle. However, accessing this information could be tedious and time-consuming for non-technical users, who might have limited or no knowledge of working with BIM software. Automating the information inquiry process can potentially address this need. This research proposes an Artificial Intelligence-based framework to facilitate accessing information in BIM models. First, the framework uses a support vector machine (SVM) algorithm to determine the user's question type. Simultaneously, it employs natural language processing (NLP) for syntactic analysis to find the main keywords of the user's question. Then it utilizes an ontology database such as IfcOWL and an NLP method (latent semantic analysis (LSA)) for a semantic understanding of the question. The keywords are expanded through the semantic relationship in the ontologies, and eventually, a final query is formed based on keywords and their expanded concepts. A Navisworks API is developed that employs the identified question type and its parameters to extract the results from BIM and display them to the users. The proposed platform also includes a speech recognition module for a more user-friendly interface. The results show that the speed of answering the questions on the platform is up to 5 times faster than the manual use by experts while maintaining high accuracy.

A latent semantic analysis method for ranking the results of human disease search engine

The human disease search engine based on the search query about disease factors (symptom, cause, position happening symptoms, i.e.,) helps users to conveniently diagnose the disease they may have anytime, anywhere. Therefore, the disease results returned by the search engine need to be accurate and ranked reasonably so that users can know which disease has the highest probability for their search query. We propose a method to arrange the returned diseases based on the latent semantic analysis (LSA) technique. This method helps to rank the disease results reasonably and meaningfully because it not only exploits the matching term frequency-inverse document frequency (TF-IDF) scores between the disease factors in the query and the disease results, but it also exploits the implicit relationship between the disease factors in the search query and the disease factors in the disease results.

Sustainable Marketing and the Role of Social Media: An Experimental Study Using Natural Language Processing (NLP)

Marketing has changed fundamentally in the new millennium. At the same time, sustainable marketing strategies have evolved to meet the challenges of environmental issues. In this study, we examined the trends in sustainable marketing strategies and the role of social media in these. Based on specific keywords per the objective, this study collected 33 published articles from the Scopus database from 1991 to 2022 (2012–2022). The KNIME (Konstanz Information Miner) and VOSviewer tools were deployed to provide detailed classification and prediction of the various trends in sustainable marketing, with a particular focus on the role of social media. The study method applied text mining and latent semantic analysis to predict the latest trends. The top three trends were Green Marketing and Consumer Behavior, Sustainable Social Media Marketing, and Influencer Social Media Marketing Practices. This NLP-based review and the clustering of research directions provide immense value to marketers and policymakers.

Testing Computational Assessment of Idea Novelty in Crowdsourcing

ABSTRACT In crowdsourcing ideation websites, companies can easily collect large amount of ideas. Screening through such volume of ideas is very costly and challenging, necessitating automatic approaches. It would be particularly useful to automatically evaluate idea novelty since companies commonly seek novel ideas. Four computational approaches were tested, based on Latent Semantic Analysis (LSA), Latent Dirichlet Allocation (LDA), term frequency – inverse document frequency (TF-IDF), and Global Vectors for Word Representation (GloVe), respectively. These approaches were used on three set of ideas and the computed idea novelty scores, along with crowd evaluation, were compared with human expert evaluation. The computational methods do not differ significantly with regard to correlation coefficients with expert ratings, even though TF-IDF-based measure achieved a correlation above 0.40 in two out of the three tasks. Crowd evaluation outperforms all the computational methods. Overall, our results show that the tested computational approaches do not match human judgment well enough to replace it.

Implementation of a Noise Filter for Grouping in Bibliographic Databases using Latent Semantic Indexing

Clustering algorithms can assist in scientific research by presenting themes related to some topics from which we can extract information more easily. However, it is common for many of these clusters to have documents that have no relevance to the topic of interest, thereby reducing the quality of the information. We can manage the reduced quality of information of clusters for a bibliographic database by dealing with noise in the semantic space that represents the relations between the grouped documents. In this work, we sustain the hypothesis of using the Latent Semantic Indexing (LSI) technique as an efficient instrument to reduce noise and promote better group quality. Using a database of 90 scientific publications from different areas, we pre-processed the documents by LSI and grouped them using six clustering algorithms. The results were significantly improved compared to our initial results that did not use LSI-based pre-processing. From the perspective of individual performance of the algorithms demonstrating the best results, CMeans was the one that got the highest average gain, with approximately 25%, followed by K-Means and SKmeans, with 17% each; PAM, with 16.5%; and EM, with 15%. The conclusion is that Latent Semantic Indexing has proven to be a helpful tool for noise reduction. We recommend its use to improve the cluster quality of bibliographic databases significantly.

Detecting Group Shilling Attacks In Recommender Systems Based On User Multi-dimensional Features And Collusive Behaviour Analysis

Abstract Group shilling attacks are more threatening than individual shilling attacks due to the collusive behaviours among group members, which pose a great challenge to the credibility of recommender systems. Detection of group shilling attacks can reduce the risk caused by such attacks and ensure the credibility of recommendations. The existing methods for detecting group shilling attacks mainly extract features from the rating patterns of users at group level to measure the shilling behaviours of groups. However, they may become ineffective with the change of attack strategy, resulting in a decrease in detection performance. Aiming at this problem, a new solution based on user multi-dimensional features and collusive behaviour analysis is presented for detecting group shilling attacks. First, we employ the information entropy and latent semantic analysis to analyse the user behavioural patterns from dimensions of item, rating, time and interest, and propose a suite of indicators to measure the anomaly behaviours of users. Second, we propose a measure based on the multi-dimensional features of users to capture the collusion of group members from the perspective of their synchronized behaviours and abnormal behaviours, and treat the groups with high collusion as candidate groups. Finally, based on the multi-dimensional features of users, we construct the user behaviour similarity matrix using Gaussian radial basis function (Gaussian-RBF) and adopt the spectral clustering algorithm to spot group shilling attackers in the candidate groups. Experiments show that the detection performance (F1-measure) of the proposed method can achieve 0.965, 0.964, 0.991 and 0.868 on the Netflix, CiaoDVD, Epinions and Amazon datasets, respectively, which is better than that of state-of-the-art methods.

How to detect propaganda from social media? Exploitation of semantic and fine-tuned language models.

Online propaganda is a mechanism to influence the opinions of social media users. It is a growing menace to public health, democratic institutions, and public society. The present study proposes a propaganda detection framework as a binary classification model based on a news repository. Several feature models are explored to develop a robust model such as part-of-speech, LIWC, word uni-gram, Embeddings from Language Models (ELMo), FastText, word2vec, latent semantic analysis (LSA), and char tri-gram feature models. Moreover, fine-tuning of the BERT is also performed. Three oversampling methods are investigated to handle the imbalance status of the Qprop dataset. SMOTE Edited Nearest Neighbors (ENN) presented the best results. The fine-tuning of BERT revealed that the BERT-320 sequence length is the best model. As a standalone model, the char tri-gram presented superior performance as compared to other features. The robust performance is observed against the combination of char tri-gram + BERT and char tri-gram + word2vec and they outperformed the two state-of-the-art baselines. In contrast to prior approaches, the addition of feature selection further improves the performance and achieved more than 97.60% recall, f1-score, and AUC on the dev and test part of the dataset. The findings of the present study can be used to organize news articles for various public news websites.

“Computational analysis on verbal fluency reveals heterogeneity in subjective language interests and brain structure”

Language is an essential higher cognitive function in humans and is often affected by psychiatric and neurological disorders. Objective measures like the verbal fluency test are often used to determine language dysfunction. Recent applications of computational approaches broaden insights into language-related functions. In addition, individuals diagnosed with a psychiatric or neurological disorder also often report subjective difficulties in language-related functions. Therefore, we investigated the association between objective and subjective measures of language functioning, on the one hand, and inter-individual structural variations in language-related brain areas, on the other hand.We performed a Latent Semantic analysis (LSA) on a semantic verbal fluency task in 101 healthy adult participants. To investigate if these objective measures are associated with a subjective one, we examined assessed subjective natural tendency of interest in language-related activity with a study-specific questionnaire. Lastly, a voxel-based brain morphometry (VBM) was conducted to reveal associations between objective (LSA) measures and structural changes in language-related brain areas.We found a positive correlation between the LSA measure cosine similarity and the subjective interest in language. Furthermore, we found that higher cosine similarity corresponds to higher gray matter volume in the right cerebellum. The results suggest that people with higher interests in language access semantic knowledge in a more organized way exhibited by higher cosine similarity and have larger gray matter volume in the right cerebellum, when compared to people with lower interests.In conclusion, we demonstrate that there is inter-individual diverseness of accessing the semantic knowledge space and that it is associated with subjective language interests as well as structural differences in the right cerebellum.

Android-IoT Malware Classification and Detection Approach Using Deep URL Features Analysis

Currently, malware attacks pose a high risk to compromise the security of Android-IoT apps. These threats have the potential to steal critical information, causing economic, social, and financial harm. Because of their constant availability on the network, Android apps are easily attacked by URL-based traffic. In this paper, an Android malware classification and detection approach using deep and broad URL feature mining is proposed. This study entails the development of a novel traffic data preprocessing and transformation method that can detect malicious apps using network traffic analysis. The encrypted URL-based traffic is mined to decrypt the transmitted data. To extract the sequenced features, the N-gram analysis method is used, and afterward, the singular value decomposition (SVD) method is utilized to reduce the features while preserving the actual semantics. The latent features are extracted using the latent semantic analysis tool. Finally, CNN-LSTM, a multi-view deep learning approach, is designed for effective malware classification and detection.

L2 and L1 semantic context indices as automated measures of lexical sophistication

Lexical frequency benchmarks have been extensively used to investigate second language (L2) lexical sophistication, especially in language assessment studies. However, indices based on semantic co-occurrence, which may be a better representation of the experience language users have with lexical items, have not been sufficiently tested as benchmarks of lexical sophistication. To address this gap, we developed and tested indices based on semantic co-occurrence from two computational methods, namely, Latent Semantic Analysis and Word2Vec. The indices were developed from one L2 written corpus (i.e., EF Cambridge Open Language Database [EF-CAMDAT]) and one first language (L1) written corpus (i.e., Corpus of Contemporary American English [COCA] Magazine). Available L1 semantic context indices (i.e., Touchstone Applied Sciences Associates [TASA] indices) were also assessed. To validate the indices, they were used to predict L2 essay quality scores as judged by human raters. The models suggested that the semantic context indices developed from EF-CAMDAT and TASA, but not the COCA Magazine indices, explained unique variance in the presence of lexical sophistication measures. This study suggests that semantic context indices based on multi-level corpora, including L2 corpora, may provide a useful representation of the experience L2 writers have with input, which may assist with automatic scoring of L2 writing.

Short text topic modelling using local and global word-context semantic correlation.

Nowadays, people use short text to portray their opinions on platforms of social media such as Twitter, Facebook, and YouTube, as well as on e-commerce websites such as Amazon and Flipkart to share their commercial purchasing experiences. Every day, billions of short texts are created worldwide in tweets, tags, keywords, search queries etc. However, this short text possesses inadequate contextual information, which can be ambiguous, sparse, noisy, remains a major challenge. State-of-the-art strategies of topic modeling such as Latent Dirichlet Allocation and Probabilistic Latent Semantic Analysis are not suitable as it contains a limited number of words in a single document. This work proposes a new model named G_SeaNMF (Gensim_SeaNMF) to improve the word-context semantic relationship by using local and global word embedding techniques. Word embeddings learned from a large corpus provide general semantic and syntactic information about words; it can guide topic modeling for short text collections as supporting information for sparse co-occurrence patterns. In the proposed model, SeaNMF (Semantics-assisted Non-negative Matrix Factorization) is incorporated with word2vec model of Gensim library to strengthen the word's semantic relationship. In this article, a short text topic modeling techniques based on DMM (Dirichlet Multinomial Mixture), self-aggregation and global word co-occurrence were explored. These are evaluated using different measures to gauge cluster coherence on real-world datasets such as Search Snippet, Biomedicine, Pascal Flickr, Tweet and TagMyNews. Empirical evaluation shows that a combination of local and global word embedding provides more appropriate words under each topic with improved outcomes.

AgroSupportAnalytics: big data recommender system for agricultural farmer complaints in Egypt

<span>The world’s agricultural needs are growing with the pace of increase in its population. Agricultural farmers play a vital role in our society by helping us in fulfilling our basic food needs. So, we need to support farmers to keep up their great work, even in difficult times such as the coronavirus disease (COVID-19) outbreak, which causes hard regulations like lockdowns, curfews, and social distancing procedures. In this article, we propose the development of a recommender system that assists in giving advice, support, and solutions for the farmers’ agricultural related complaints (or queries). The proposed system is based on the latent semantic analysis (LSA) approach to find the key semantic features of words used in agricultural complaints and their solutions. Further, it proposes to use the support vector machine (SVM) algorithm with Hadoop to classify the large agriculture dataset over Map/Reduce framework. The results show that a semantic-based classification system and filtering methods can improve the recommender system. Our proposed system outperformed the existing interest recommendation models with an accuracy of 87%.</span>

Discussions about the Russian Revolution of 1917 on the Web: methodological approaches to the study of web forums as a historical source

The subject of the study is a methodology for analyzing the electronic content of social networks (forums) as a historical source. The discussion of the revolution of 1917 during the centenary of this historical event was used as a material for analysis. The aim of the study was to test approaches to the methodology of working with large arrays of online texts, and the possible combination of two approaches to working with online texts - quantitative analysis tools (distant reading) and traditional methods of working with historical text (slow reading). As part of the "distant reading", thematic modeling is used using the LDA (latent Dirichlet placement) and LSA (latent semantic analysis) algorithm in the R programming environment in the R studio program (version 4.2.1). During the "slow reading" we analyze the entire volume of the text directly.The novelty of the research lies in the application of thematic modeling to sources in the R programming environment in conjunction with classical methods of analyzing historical texts. Within the framework of the study, a methodology for analyzing the content of social networks (forums) has been tested, focused on substantial arrays of text that are physically impossible to read in full or at least in a significant part, using exclusively traditional means of interaction of the researcher with the corpus of sources. A step-by-step research algorithm is proposed, in which the researcher needs to analyze the text by "distant reading" methods, identifying the topics of texts consisting of terms (words). Then, using these keywords, you should find the relevant text fragments in which the identified topic was discussed most actively, and analyze the fragments in more detail using traditional methods of working with a text source. A possible way to improve the quality of identifying topics necessary for the researcher in social networks and forums by the LDA algorithm is proposed, namely, preliminary splitting of a large text and subsequent analysis of fragments by the LDA method as separate documents.

English Language Learning via YouTube: An NLP-Based Analysis of Users’ Comments

Online teaching and learning has been beneficial in facilitating learning of English as a foreign language (EFL). In online EFL learning, YouTube is one of the most utilized information and communication technology (ICT) tools because of its inherent features that make it a unique environment for learners and educators. Many interesting aspects of YouTube-based learning can be beneficial in supplementing conventional classroom methods, and, therefore, such aspects must be identified. Previous scholarly work aimed at improving YouTube learning environment was predominantly conducted manually by gathering learners’ impressions through interviews and questionnaires to analyze the differences between YouTube- and classroom-based EFL learning. However, such methods are tedious and time-consuming and can lead to results that are of less generalizable implications. User comments on YouTube channels are useful in identifying such aspects, as they present a wealth of information related to the quality of the content provided, challenges the targeted audience faces, and areas of potential improvement. Therefore, in our current study, YouTube API is used to collect the comments of three randomly selected and popular YouTube channels. Following a data cleaning process, people’s sentiments about EFL learning were first identified via a TextBlob method. Second, the automated latent semantic analysis (LSA) method of topic finding was used to collect global and open-ended topics of discussion on YouTube-based EFL learning. Users’ sentiments on the most popular topics of discussion are discussed in this paper. Further, based on the results, hypothetical findings on YouTube EFL learning are provided as recommendation for future content, including more variety of the content covered, introduction of the meanings and punctuation following words, the design of the course such that it addresses a multinational audience of any age, and targeted teaching of each variety of English, such as British and American. We also make suggestions for learners of English who wish to utilize online and offline learning, which include finding the course of interest first based on one’s needs which can be discussed with a tutor or any English teacher to optimize the learning experience, participating in fearless educator–learner interaction and engagement, and asking other EFL learners for their previous experiences with learning online in order for the learner to maximize benefit.

Contextualizing automated writing evaluation: a case of english for specific purposes writings

With the developments of technology, writing evaluation and feedback are supported by Natural Language Processing (NLP) and Latent Semantic Analysis (LSA). The immersion of NLP and LSA in Writing contributes to the development of Automatic Writing Evaluation (AWE), a system that compares a text to a large database of the writing of the same genre and supports an evaluation process that is independent of the human rater. The potential of T.E.R.A's reach in evaluation became one of the bases of this research. This research intends to compare the evaluation process of human raters to T.E.R.A. and find out; a) the figuration of assessment criteria used in the evaluation process, and b) which elements rates are higher, the same, or lower. The result of this study shall be a point of departure to establish AWE which is suitable to the context of English as a Foreign Language being taught as a compulsory course at Bina Nusantara University.

Visualizing design project team and individual progress using NLP: a comparison between latent semantic analysis and Word2Vector algorithms

Abstract Design has always been seen as an inherently human activity and hard to automate. It requires a lot of traits that are seldom attributable to machines or algorithms. Consequently, the act of designing is also hard to assess. In particular in an educational context, the assessment of progress of design tasks performed by individuals or teams is difficult, and often only the outcome of the task is assessed or graded. There is a need to better understand, and potentially quantify, design progress. Natural Language Processing (NLP) is one way of doing so. With the advancement in NLP research, some of its models are adopted into the field of design to quantify a design class performance. To quantify and visualize design progress, the NLP models are often deployed to analyze written documentation collected from the class participants at fixed time intervals through the span of a course. This paper will explore several ways of using NLP in assessing design progress, analyze its advantages and shortcomings, and present a case study to demonstrate its application. The paper concludes with some guidelines and recommendations for future development.

A Novel Spam Classification System for E-Mail Using a Gradient Fuzzy Guideline-Based Spam Classifier (GFGSC)

Spam messages have increased dramatically in recent years even as the number of email clients has grown. Email has already become a valuable way of communicating because it saves time and effort. However, numerous emails contain unwelcome content known as spam as a result of social platforms and advertisements. Despite the fact that many techniques have already been created for spam mails categorization, none of them achieves 100 percent efficiency in analyzing spam messages. So, in this research, we propose a novel Gradient Fuzzy Guideline-based Spam Classifier (GFGSC) for classifying the spam e-mails as spam or non-spam. This research uses four types of datasets and these datasets are pre-processed using normalization. Then the set of data can be extracted using Principal Component Analysis (PCA) and Latent Semantic Analysis (LSA) techniques. The aspects are selected using Information Gain (IG) and Chi-Square (ChS) techniques. And the GFGSC classifier can be used for classifying the data as spam or non-spam with better effectiveness. Finally, the performances are examined and these metrics are matched with the existing approaches. The results are obtained using the MATLAB tool.