LSA Method Research Articles

A Sentiment Analysis and Topic Modelling of The Socio-Economic Registration 2022

Socio-Economic Registration or Regsosek is an activity of Statistics Indonesia (BPS) that aims to collect data related to the profile, social and economic conditions, and welfare levels of all residents in 514 regencies/cities in Indonesia. One indicator of the success of Regsosek 2022 is the response and opinion from the community regarding the activity. The response and opinion can provide an overview of the implementation of Regsosek 2022 so that the picture can be used as a lesson learned to carry out the following population data collection. This study uses several methods to analyze the results of community responses and opinions on Regsosek activities, especially on Twitter social media. The method used in this research is sentiment analysis classification with four techniques: Naïve Bayes, Nearest Centroid, K-Nearest Neighbors, and Support Vector Machine. Then, the performance of the four techniques will be compared. In addition, the topic modeling method will also be used with two techniques, namely Latent Semantic Analysis and Latent Dirichlet Allocation. Data is collected using web scraping techniques. The results obtained from the sentiment analysis classification are that the Nearest Centroid method provides the best results with a relatively high and balanced f1-score value in positive and negative sentiments, which are 59% and 66%, respectively. Moreover, LDA modeling results are better than the LSA method for topic modeling results.

The Use of Natural Language Processing for Computer-Aided Diagnostics and Monitoring of Body Image Perception in Patients with Cancers.

Head and neck cancers (H&NCs) constitute a significant part of all cancer cases. H&NC patients experience unintentional weight loss, poor nutritional status, or speech disorders. Medical interventions affect appearance and interfere with patients' self-perception of their bodies. Psychological consultations are not affordable due to limited time. We used NLP to analyze the basic emotion intensity, sentiment about one's body, characteristic vocabulary, and potential areas of difficulty in free notes. The emotion intensity research uses the extended NAWL dictionary developed using word embedding. The sentiment analysis used a hybrid approach: a sentiment dictionary and a deep recursive network. The part-of-speech tagging and domain rules defined by a psycho-oncologist determine the distinct language traits. Potential areas of difficulty were analyzed using the dictionaries method with word polarity to define a given area and the presentation of a note using bag-of-words. Here, we applied the LSA method using SVD to reduce dimensionality. A total of 50 cancer patients requiring enteral nutrition participated in the study. The results confirmed the complexity of emotions in patients with H&NC in relation to their body image. A negative attitude towards body image was detected in most of the patients. The method presented in the study appeared to be effective in assessing body image perception disturbances, but it cannot be used as the sole indicator of body image perception issues. The main problem in the research was the fairly wide age range of participants, which explains the potential diversity of vocabulary. The combination of the attributes of a patient's condition, possible to determine using the method for a specific patient, can indicate the direction of support for the patient, relatives, direct medical personnel, and psycho-oncologists.

Підхід до виявлення аномалій в потоках тектових даних

Збільшення потоків інформації веде за собою необхідність розробки спеціалізованих інтелектуальних засобів та методів для обробки величезної кількості даних. Так популярність соціальних мереж, різного роду месенджерів вимагає створення спеціалізованих засобів для обробки потоку текстових повідомлень. Дане дослідження присвячене дослідженню та розробці методу виявлення аномальних елементів в потоках текстових даних. Особлива увага присвячена підтримці україномовних текстів.

PAPR Reduction for O-OFDM UOWC System

Abstract In this paper, a combination of Tone Reservation-Signal to Clipping noise Ratio (TR-SCR) and Companding schemes is proposed for underwater wireless optical communication (UOWC), in order to solve the problem of high peak-to-average power ratio (PAPR) of underwater Optical-Orthogonal Frequency Division Multiplexing (O-OFDM) system. Two optical modulation schemes, namely Direct Current-biased Optical OFDM (DCO-OFDM) and Asymmetrically Clipped Optical OFDM (ACO-OFDM), are adopted separately to achieve intensity modulation/direct detection over the seawater channel, with a distance of 100 meters of wireless optical transition to obtain the system bit error rate (BER). In the TR-SCR scheme, only the maximum amplitude signal is selected to reduce the complexity of the algorithm. The LSA method is introduced in the TR-SCR scheme for iteratively updating the signal to speed up the convergence. Finally, with the combination of the Companding scheme, the PAPR performance is significantly improved. The PAPR of the underwater optical OFDM system is reduced to 3.85 dB with its BER still satisfying the UOWC system requirements.

LATENT-SEMANTIC ANALYSIS, SOCIAL NETWORKS AND NON-STRUCTURED DATA: INTERACTION METHOD

This article examines the method of latent-semantic analysis, its advantages, disadvantages, and the possibility of further transformation for use in arrays of unstructured data, which make up most of the information that Internet users deal with. To extract context-dependent word meanings through the statistical processing of large sets of textual data, an LSA method is used, based on operations with numeric matrices of the word-text type, the rows of which correspond to words, and the columns of text units to texts. The integration of words into themes and the representation of text units in the theme space is accomplished by applying one of the matrix expansions to the matrix data: singular decomposition or factorization of nonnegative matrices. The results of LSA studies have shown that the content of the similarity of words and text is obtained in such a way that the results obtained closely coincide with human thinking. Based on the methods described above, the author has developed and proposed a new way of finding semantic links between unstructured data, namely, information on social networks. The method is based on latent-semantic and frequency analyzes and involves processing the search result received, splitting each remaining text (post) into separate words, each of which takes the round in n words right and left, counting the number of occurrences of each term, working with a pre-created semantic resource (dictionary, ontology, RDF schema, ...). The developed method and algorithm have been tested on six well-known social networks, the interaction of which occurs through the ARI of the respective social networks. The average score for author's results exceeded that of their own social network search. The results obtained in the course of this dissertation can be used in the development of recommendation, search and other systems related to the search, rubrication and filtering of information.

A novel peak-to-average power ratio reduction scheme via tone reservation in OFDM systems

In this paper, a novel signal-to-clipping noise ratio and least squares approximation tone reservation scheme (SCR-LSA TR) is proposed to reduce the peak-to-average power ratio for orthogonal frequency division multiplexing systems. During the SCR procedure, only the element with the maximal amplitude is picked for processing, which not only decreases the algorithm complexity, but also helps to overcome the BER deterioration. With the LSA method, the amplitude of the peak-cancelling signals can approximate to that of the original clipping noise as much as possible. Through the combination of the optimization factor in the LSA method, the classic SCR method can achieve better PAPR reduction with faster convergence. Simulation results show that the proposed SCR-LSA TR scheme has less in-band distortion and smaller out-of-band spectral radiation. The BER of the proposed scheme shows a better performance especially under the 16-QAM over the additive white Gaussian noise channel.

Development of a computer system for generating semantic template of a group of documents by using latent semantic analysis

The CS was developed by means of the Python programming language to generate a semantic template of a group of documents by the LSA method. The system contains eight software modules, each performs one stage of the LSA. The control module of the frequency word-document matrix and the measuring module of semantic distance between the template documents are unique. Adjustment of CS to the contents and structure of the document templates is performed by changing a set of modules. According to the research, the frequency matrix normalization enhances the resolution of the semantic template generated by using the LSA. It is proved that the removal of individual words improves the resolution of the generated semantic template and does not affect the semantic content. Application of semantic proximity of documents, the cosine of the difference of angles between the vector of a group of basic words and vectors of documents for evaluation allows increasing the resolution of the generated semantic template. To ensure the continuity of the LSA, the module of the frequency matrix analysis for compliance of excess (or equality) of the number of words over the number of documents was introduced in the CS. In the event of a mismatch, the module starts over the LSA process with a new set of words and documents after removal of the inappropriate document and related words.

SEMCON

This paper presents a novel concept enrichment objective metric combining contextual and semantic information of terms extracted from the domain documents. The proposed metric is called SEMCON which stands for semantic and contextual objective metric. It employs a hybrid learning approach utilizing functionalities from statistical and linguistic ontology learning techniques. The metric also introduced for the first time two statistical features that have shown to improve the overall score ranking of highly relevant terms for concept enrichment. Subjective and objective experiments are conducted in various domains. Experimental results (F1) from computer domain show that SEMCON achieved better performance in contrast to tf*idf, and LSA methods, with 12.2%, 21.8%, and 24.5% improvement over them respectively. Additionally, an investigation into how much each of contextual and semantic components contributes to the overall task of concept enrichment is conducted and the obtained results suggest that a balanced weight gives the best performance.

SCESS: a WFSA-based automated simplified chinese essay scoring system with incremental latent semantic analysis

AbstractWriting in language tests is regarded as an important indicator for assessing language skills of test takers. As Chinese language tests become popular, scoring a large number of essays becomes a heavy and expensive task for the organizers of these tests. In the past several years, some efforts have been made to develop automated simplified Chinese essay scoring systems, reducing both costs and evaluation time. In this paper, we introduce a system called SCESS (automated Simplified Chinese Essay Scoring System) based on Weighted Finite State Automata (WFSA) and using Incremental Latent Semantic Analysis (ILSA) to deal with a large number of essays. First, SCESS uses ann-gram language model to construct a WFSA to perform text pre-processing. At this stage, the system integrates a Confusing-Character Table, a Part-Of-Speech Table, beam search and heuristic search to perform automated word segmentation and correction of essays. Experimental results show that this pre-processing procedure is effective, with a Recall Rate of 88.50%, a Detection Precision of 92.31% and a Correction Precision of 88.46%. After text pre-processing, SCESS uses ILSA to perform automated essay scoring. We have carried out experiments to compare the ILSA method with the traditional LSA method on the corpora of essays from the MHK test (the Chinese proficiency test for minorities). Experimental results indicate that ILSA has a significant advantage over LSA, in terms of both running time and memory usage. Furthermore, experimental results also show that SCESS is quite effective with a scoring performance of 89.50%.

Wireless Sensor Network Localization Based on Semi-Supervised Local Subspace Alignment

A semi-supervised local subspace alignment (SSLSA) algorithm for wireless sensor networks localization is presented. SSLSA method has the following features: 1) only using pair-wise distance between each node and its neighbors within a certain communication range, 2) aligning subspaces with the constraints of anchor nodes thus avoiding the accumulation of error, and 3) obtaining absolute positions of all sensor nodes directly without any other transformation process. The localization performance of this method is compared to the MDS-MAP(P, R) and LSA methods.

Evaluation de l'analyse sémantique latente et du modèle vectoriel standard appliqués à la langue arabe

In the objective of a possible performance improvement of the Arabic information retrieval systems, we propose to introduce the latent semantic analysis method to cure the problems arising from the vector- space model. The present contribution describes how linguistic processing and weighting schemes could improve the LSA method, and the comparison between the vector-space model and LSA approach, which aim to reduce the index term number of an Arabic corpus specialized in the environment field. The results of our experiments show clearly a positive influence of the linguistic processing and weighting schemes, and LSA improvement compared to the vector-space model.

Theoretical prediction of prefered bonding sites: a critical analysis based on a model system

We develop a critical analysis of different procedures found in the literature to theoretically predict preferred bonding sites in chemisorption. Geometrical and edge effects, which appear in the calculation of adatom-surface potential energy, paralel to the surface, are investigated within the embedding LSA and finite-cluster methods. A systematic study of convergence for different adsorbate sites is performed on the chemisorption of H on a model tungsten (100) surface. By using arguments based strictly on the convergence analysis and on the N-variational character of the LSA method, we show that, due to embedding, the latter formalism eliminates geometrical (shape) effects of the finite- N representation; it is shown to be suitable to represent in general the potential energy curve as a function of parallel coordinates. It is also shown that finite-cluster models may lead to erroneous interpretation. On the other hand, certain kinds of cluster geometries present a smooth convergence for the binding energy, contrary to other geometries investigated in previous works. This fact leaded to a suggestion for the best procedure to compare cluster binding energies of different sites in the cubium model.