Application Of Feature Extraction Research Articles

Powering piezoelectric frequency up-converter with rotary magnetic forces for torque-sensing application

A piezoelectric frequency up-energy harvester (PFU-EH) utilizing an SECE circuit (synchronized electric charge extraction) is developed for torque-sensing application in rotating machinery. Development of an explicit theoretical model for the system with the PFU-EH, considering phase angle shifts and magnetic impulsive forces is proposed to formulate the frequency response equations for voltage and power outputs using Fourier analysis. In this study, harvesting electrical energy can be achieved through the magnetic force interaction between the cantilevered piezoelectric beam, which carries two attached magnets, and the compound rotating disk structure, also carrying two attached magnets. The compound rotating disk structure, composed of a metal ring (outer disk) and a flywheel (inner disk) connected by a spring system, can create a phase angle caused by an applied load. As a result, mechanical torque is formed for inducing magnetic excitation to the PFU-EH and subsequently generating an AC voltage signal. The functionality of the AC voltage harvested by the PFU-EH can be further leveraged for two primary benefits. First, it can be converted into DC voltage output using the SECE circuit, enhancing the effectiveness of the electric signal without requiring load matching under different modal vibrations. Second, the SECE harvesting voltage can simultaneously function as a sensing signal for detecting mechanical torque itself. The sensing signal detection process is further analyzed by applying feature extraction from the spectrogram and utilizing a 2D-CNN for auxiliary segment identification. Finally, the experimental results demonstrate reasonable agreement with the predicted outcomes.

Quantum-enhanced hybrid feature engineering in thoracic CT image analysis for state-of-the-art nodule classification: an advanced lung cancer assessment

The intricate nature of lung cancer treatment poses considerable challenges upon diagnosis. Early detection plays a pivotal role in mitigating its escalating global mortality rates. Consequently, there are pressing demands for robust and dependable early detection and diagnostic systems. However, the technological limitations and complexity of the disease make it challenging to implement an efficient lung cancer screening system. AI-based CT image analysis techniques are showing significant contributions to the development of computer-assisted detection (CAD) systems for lung cancer screening. Various existing research groups are working on implementing CT image analysis systems for assessing and classifying lung cancer. However, the complexity of different structures inside the CT image is high and comprehension of significant information inherited by them is more complex even after applying advanced feature extraction and feature selection techniques. Traditional and classical feature selection techniques may struggle to capture complex interdependencies between features. They may get stuck in local optima and sometimes require additional exploration strategies. Traditional techniques may also struggle with combinatorial optimization problems when applied to a prominent feature space. This paper proposed a methodology to overcome the existing challenges by applying feature extraction using Vision Transformer (FexViT) and Feature selection using the Quantum Computing based Quadratic unconstrained binary optimization (QC-FSelQUBO) technique. This algorithm shows better performance when compared with other existing techniques. The proposed methodology showed better performance as compared to other existing techniques when evaluated by applying necessary output measures, such as accuracy, Area under roc (receiver operating characteristics) curve, precision, sensitivity, and specificity, obtained as 94.28%, 99.10%, 96.17%, 90.16% and 97.46%. The further advancement of CAD systems is essential to meet the demand for more reliable detection and diagnosis of cancer, which can be addressed by leading the proposed quantum computation and growing AI-based technology ahead.

Sentiment analysis of Arabic social media texts: A machine learning approach to deciphering customer perceptions

Sentiment analysis (SA) is a subfield of artificial intelligence that entails natural language processing. This has become increasingly significant because it discerns the emotional tone of reviews, categorising them as positive, neutral, or negative. In the highly competitive coffee industry, understanding customer sentiment and perception is paramount for businesses seeking to optimise their product offerings. Traditional methods of market analysis often fall short of capturing the nuanced views of consumers, necessitating a more sophisticated approach to sentiment analysis. This research is motivated by the need for a nuanced understanding of customer sentiments across various coffee products, enabling companies to make informed decisions regarding product promotion, improvement, and discontinuation. However, sentiment analysis faces a challenge when it comes to analysing Arabic text due to the language's extraordinarily complex inflectional and derivational morphology. Consequently, to address this challenge, we have developed a new method designed to improve the precision and effectiveness of Arabic sentiment analysis, specifically focusing on understanding customer opinions about various coffee products on social media platforms like Twitter. We gathered 10,646 various coffee products' Twitter reviews and applied feature extraction techniques using the term frequency-inverse document frequency (TF-IDF) and minimum redundancy maximum relevance (MRMR). Subsequently, we performed sentiment analysis using four supervised learning algorithms: k-nearest neighbor, support vector machine, decision tree, and random forest. All the classification statements derived in the analysis were aggregated via ensemble learning to convey the final results. Our results demonstrated an increase in prediction accuracy, with our method achieving over 95.95% accuracy in the Hard voting and soft voting at 94.51 %.

Reimagining Literary Analysis: Utilizing Artificial Intelligence to Classify Modernist French Poetry

Aligned with global Sustainable Development Goals (SDGs) and multidisciplinary approaches integrating AI with sustainability, this research introduces an innovative AI framework for analyzing Modern French Poetry. It applies feature extraction techniques (TF-IDF and Doc2Vec) and machine learning algorithms (especially SVM) to create a model that objectively classifies poems by their stylistic and thematic attributes, transcending traditional subjective analyses. This work demonstrates AI’s potential in literary analysis and cultural exchange, highlighting the model’s capacity to facilitate cross-cultural understanding and enhance poetry education. The efficiency of the AI model, compared to traditional methods, shows promise in optimizing resources and reducing the environmental impact of education. Future research will refine the model’s technical aspects, ensuring effectiveness, equity, and personalization in education. Expanding the model’s scope to various poetic styles and genres will enhance its accuracy and generalizability. Additionally, efforts will focus on an equitable AI tool implementation for quality education access. This research offers insights into AI’s role in advancing poetry education and contributing to sustainability goals. By overcoming the outlined limitations and integrating the model into educational platforms, it sets a path for impactful developments in computational poetry and educational technology.

Retinal Disease Diagnosis Using Deep Learning on Ultra-Wide-Field Fundus Images.

Ultra-wide-field fundus imaging (UFI) provides comprehensive visualization of crucial eye components, including the optic disk, fovea, and macula. This in-depth view facilitates doctors in accurately diagnosing diseases and recommending suitable treatments. This study investigated the application of various deep learning models for detecting eye diseases using UFI. We developed an automated system that processes and enhances a dataset of 4697 images. Our approach involves brightness and contrast enhancement, followed by applying feature extraction, data augmentation and image classification, integrated with convolutional neural networks. These networks utilize layer-wise feature extraction and transfer learning from pre-trained models to accurately represent and analyze medical images. Among the five evaluated models, including ResNet152, Vision Transformer, InceptionResNetV2, RegNet and ConVNext, ResNet152 is the most effective, achieving a testing area under the curve (AUC) score of 96.47% (with a 95% confidence interval (CI) of 0.931-0.974). Additionally, the paper presents visualizations of the model's predictions, including confidence scores and heatmaps that highlight the model's focal points-particularly where lesions due to damage are evident. By streamlining the diagnosis process and providing intricate prediction details without human intervention, our system serves as a pivotal tool for ophthalmologists. This research underscores the compatibility and potential of utilizing ultra-wide-field images in conjunction with deep learning.

Ultra-short-term solar forecasting with reduced pre-acquired data considering optimal heuristic configurations of deep neural networks

<abstract> <p>Forecasting solar irradiance, particularly Global Horizontal Irradiance (GHI), has drawn much interest recently due to the rising demand for renewable energy sources. Many works have been proposed in the literature to forecast GHI by incorporating weather or environmental variables. Nevertheless, the expensive cost of the weather station hinders obtaining meteorological data, posing challenges in generating accurate forecasting models. Therefore, this work addresses this issue by developing a framework to reliably forecast the values of GHI even if meteorological data are unavailable or unreliable. It achieves this by leveraging lag observations of GHI values and applying feature extraction capabilities of the deep learning models. An ultra-short-term GHI forecast model is proposed using the Convolution Neural Network (CNN) algorithm, considering optimal heuristic configurations. In addition, to assess the efficacy of the proposed model, sensitivity analysis of different input variables of historical GHI observations is examined, and its performance is compared with other commonly used forecasting algorithm models over different forecasting horizons of 5, 15, and 30 minutes. A case study is carried out, and the model is trained and tested utilizing real GHI data from solar data located in Riyadh, Saudi Arabia. Results reveal the importance of employing historical GHI data in providing precise forecasting outcomes. The developed CNN-based model outperformed in ultra-short-term forecasting, showcasing average root mean square error results across different forecasting horizons: 2.262 W/m<sup>2</sup> (5min), 30.569 W/m<sup>2</sup> (15min), and 54.244 W/m<sup>2</sup> (30min) across varied day types. Finally, the findings of this research can permit GHI to be integrated into the power grid and encourage the development of sustainable energy systems.</p> </abstract>

Identifying active retrogressive thaw slumps from ArcticDEM

Permafrost degradation in the Arctic is accelerating and affects northern communities, ecosystems, and global soil carbon storage. However, the extent, distribution, and rates of permafrost degradation in the pan-Arctic remain unknown, contributing to the challenges of mapping and monitoring it in a harsh and remote environment. We applied feature extraction, deep learning, and crowdsourcing to an open-access, high-resolution (2 m), and multi-temporal digital elevation models (i.e. ArcticDEM), to identify retrogressive thaw slumps (RTSs), a dynamic form of permafrost degradation widespread across the Arctic. Specifically, we (1) developed an automated pipeline to process approximately 200 TB of ArcticDEM data; (2) designed feature extractors (pixel-wise elevation differences, polygons of elevation reductions, lines of narrow-steep slopes, and RTS headwall lines) to identify slumps; (3) trained a super-efficient object detection algorithm based on deep learning (YOLOv4) and used it to locate RTSs from composite imagery derived from the ArcticDEM; (4) combined the extracted features and the bounding boxes output by YOLOv4 to obtain mapping results at a manageable level; and (5) developed a crowdsourcing system and invited volunteers to validate and refine the results. The final map included 2494 RTSs (actively expanding during ArcticDEM observation) across the Arctic. The results also show that (1) it is necessary to combine the extracted features and deep learning to remove many false positives in the scenario with limited training data, but large regions to map; (2) some hotspots of RTSs need further and detailed investigation, including an RTS cluster in Greenland; (3) the crowdsourcing system is useful for the validation of a large dataset, but responses to this work were limited, possibly because RTSs are a quite specific topic that not many people are familiar with. The results likely miss many RTSs due to the limitation in the method for identifying RTSs and uncertainties as well as short observation periods of ArcticDEM at many locations. This study provides data and serves as a starting point to develop a global inventory and better understand permafrost thaw in the pan-Arctic using very high resolution remote sensing.

Dimensionality reduction and features visual representation based on conditional probabilities applied to activity classification

A large part of the information emitted by contemporary technological devices comes in the form of time series. The massive commercialization of these kinds of devices has made the study of time series feature extraction techniques acquire a vital relevance in last years. Two main things are essential when applying feature extraction techniques to time series: to reduce the dimensionality so it occupies the least amount of storage memory possible, and to make features that contain the relevant information regarding the nature of the data set and the goals to be achieved. For this purpose, we propose in this work a brand new technique called the State Changes Representation for Time Series (SCRTS), which relies on the relevant data associated with the conditional probabilities of the time series (also known in the literature as Markov model’s features), and the distribution of its values. This method is length-independent, which means that we can apply it to time series of different dimensions obtaining the same number of features for each one. Also, it provides a visual representation of the input data, so it is possible to interpret what makes a certain time series different from the other. After explaining how it works, we apply it to 3 different wearable accelerometer data sets. This algorithm reduces the original dimension of the time series considerably (in the best case from 5499 values to 31), having a good performance in the classification results (in the best chance with an accuracy of 98%).

Comparison of Artificial Decision Techniques for Detection of Sarcastic News Headlines

Newspapers are a rich informational source. A headline of an article sparks an interest in the reader. So, news providing agencies tend to create catchy headlines to attract the reader's attention onto them, and this is how sarcasm manages to find its way into news headlines. Sarcasm employs the use of words that carry opposite meaning with respect to what needs to be conveyed. This leads to the need of developing methods by which we can correctly predict whether a piece of text, or news for that matter, truthfully means what it says or is simply being sarcastic about it. Here, the authors have used a dataset containing 55,329 tuples consisting of news headlines from The Onion and the Huffington Post, which was taken from Kaggle, on which they applied feature extraction techniques such as Count Vectorizer, TF-IDF, Hashing Vectorizer, and Global Vectorizer (GloVe). Then they applied seven classifiers on the obtained dataset. The experimental results showed that the highest accuracies among the ML models were 81.39% for LR model with Count Vectorizer, 79.2% for LR model with TF-IDF Vectorizer, and 78% for SVM model with Count Vectorizer. They also obtained the best accuracy of 90.7% using the Bi-LSTM Deep Learning Model. They have trained the seven models and compared them based on their respective accuracies and F1-Scores.

Review of age and gender detection methods based on handwriting analysis

AbstractHandwriting recognition and analysis has been an active area of research in the last two decades. Handwriting analysis is being studied in various fields of science, such as graphology, neurology, psychology, and computer science. Furthermore, automated handwriting analysis has several applications, including forensic, security, medical, and disease prediction. This paper presents the most recent handwriting analysis techniques and advancements available in the literature for age and gender classification/detection. Different steps, including feature extraction and classification, frequently used in the literature for age and gender detection, are discussed, and the presented works are classified according to the applied feature extraction and classification methods. The online and offline benchmark databases are also reviewed. We used a text mining technique to perform a quantitative content analysis of the presented research and better understand the co-occurrence network diagrams of age and gender classification/detection. This study is a valuable resource that provides new research directions to students and researchers interested in this field for further research and investigation.

Innovations in Urdu Sentiment Analysis Using Machine and Deep Learning Techniques for Two-Class Classification of Symmetric Datasets

Many investigations have performed sentiment analysis to gauge public opinions in various languages, including English, French, Chinese, and others. The most spoken language in South Asia is Urdu. However, less work has been carried out on Urdu, as Roman Urdu is also used in social media (Urdu written in English alphabets); therefore, it is easy to use it in English language processing software. Lots of data in Urdu, as well as in Roman Urdu, are posted on social media sites such as Instagram, Twitter, Facebook, etc. This research focused on the collection of pure Urdu Language data and the preprocessing of the data, applying feature extraction, and innovative methods to perform sentiment analysis. After reviewing previous efforts, machine learning and deep learning algorithms were applied to the data. The obtained results were compared, and hybrid methods were also recommended in this research, enabling new avenues to conduct Urdu language data sentiment analysis.

Improved Spatiotemporal Framework for Human Activity Recognition in Smart Environment.

The rapid development of microsystems technology with the availability of various machine learning algorithms facilitates human activity recognition (HAR) and localization by low-cost and low-complexity systems in various applications related to industry 4.0, healthcare, ambient assisted living as well as tracking and navigation tasks. Previous work, which provided a spatiotemporal framework for HAR by fusing sensor data generated from an inertial measurement unit (IMU) with data obtained by an RGB photodiode for visible light sensing (VLS), already demonstrated promising results for real-time HAR and room identification. Based on these results, we extended the system by applying feature extraction methods of the time and frequency domain to improve considerably the correct determination of common human activities in industrial scenarios in combination with room localization. This increases the correct detection of activities to over 90% accuracy. Furthermore, it is demonstrated that this solution is applicable to real-world operating conditions in ambient light.

Integration of Low-Cost Digital Tools for Preservation of a Sustainable Agriculture System

This work presents an electronic sensing approach composed of a pair of Physical–Chemical and Imaging modules to preserve an aquaponic system. These modules offer constant measurements of the physical–chemical characteristics within the fish tank and the grow bed, and an indication of the health of the growing plants through image processing techniques. This proposal is implemented in a low-cost computer, receiving measurements from five sensors, including a camera, and processing the signals using open-source libraries and software. Periodic measurements of the temperature, water level, light, and pH within the system are collected and shared to a cloud platform that allows their display in a dashboard, accessible through a web page. The health of the vegetables growing in the system is estimated by analyzing visible and infrared spectra, applying feature extraction, and computing vegetation indices. This work provides a low-cost solution for preserving sustainable urban farming systems, suitable for new farming communities.

A Novel Road Maintenance Prioritisation System Based on Computer Vision and Crowdsourced Reporting

The maintenance of critical infrastructure is a costly necessity that developing countries often struggle to deliver timely repairs. The transport system acts as the arteries of any economy in development, and the formation of potholes on roads can lead to injuries and the loss of lives. Recently, several countries have enabled pothole reporting platforms for their citizens, so that repair work data can be centralised and visible for everyone. Nevertheless, many of these platforms have been interrupted because of the rapid growth of requests made by users. Not only have these platforms failed to filter duplicate or fake reports, but they have also failed to classify their severity, albeit that this information would be key in prioritising repair work and improving the safety of roads. In this work, we aimed to develop a prioritisation system that combines deep learning models and traditional computer vision techniques to automate the analysis of road irregularities reported by citizens. The system consists of three main components. First, we propose a processing pipeline that segments road sections of repair requests with a UNet-based model that integrates a pretrained Resnet34 as the encoder. Second, we assessed the performance of two object detection architectures—EfficientDet and YOLOv5—in the task of road damage localisation and classification. Two public datasets, the Indian Driving Dataset (IDD) and the Road Damage Detection Dataset (RDD2020), were preprocessed and augmented to train and evaluate our segmentation and damage detection models. Third, we applied feature extraction and feature matching to find possible duplicated reports. The combination of these three approaches allowed us to cluster reports according to their location and severity using clustering techniques. The results showed that this approach is a promising direction for authorities to leverage limited road maintenance resources in an impactful and effective way.

Big Data analytics for Advanced Viticulture

Big data analytics involve systematic approach to find hidden patterns to help the organization grow from large volume and variety of data. In recent years big data analytics is widely used in the agricultural domain to improve yield. Viticulture (the cultivation of grapes) is one of the most lucrative farming in India. It is a subdivision of horticulture and is the study of wine growing. The demand for Indian Wine is increasing at about 27% each year since the 21st century and thus more and more ways are being developed to improve the quality and quantity of the wine products. In this paper, we focus on a specific agricultural practice as viticulture. Weather forecasting and disease detection are the two main research areas in precision viticulture. Leaf disease detection as a part of plant pathology is the key research area in this paper. It can be applied on vineyards of India where farmers are bereft of the latest technologies. Proposed system architecture comprises four modules: Data collection, data preprocessing, classification and visualization. Database module involve grape leaf dataset, consists of healthy images combined with disease leaves such as Black measles, Black rot, and Leaf blight. Models have been implemented on Apache Hadoop using map reduce programming framework. It apply feature extraction to extract various features of the live images and classification algorithm with reduced computational complexity. Gray Level Co-occurrence Matrix (GLCM) followed by K-Nearest Neighborhood (KNN) algorithm. System also recommends the necessary steps and remedies that the viticulturists can take to assure that the grapes can be salvaged at the right time and in the right manner based on classification results. Overall system will help Indian viticulturists to improve the harvesting process. Accuracy of the model is 72% and it can be increased as a future work by including deep learning with time series grape leaf images.

Application research of pulse signal physiology and pathology feature mining in the field of disease diagnosis

This experiment is based on the principle of traditional Chinese medicine (TCM) pulse diagnosis, the human pulse signal collected by the sensor is organized into a dataset, and the algorithms are designed to apply feature extraction. After denoising, smoothing and eliminating baseline drift of the photoelectric sensors pulse data of several groups of subjects, we designed three algorithms to describe the difference between the two-dimensional images of the pulse data of normal people and patients with chronic diseases. Convert the calculated feature values into multi-dimensional arrays, enter the decision tree (DT) to balance the differences in human physiological conditions, then train in the support vector machine kernel method (SVM-KM) classifier. Experimental results show that the application of these feature mining algorithms to disease detection greatly improves the reliability of TCM diagnosis.

Robust imaging habitat computation using voxel-wise radiomics features

Tumor heterogeneity has been postulated as a hallmark of treatment resistance and a cure constraint in cancer patients. Conventional quantitative medical imaging (radiomics) can be extended to computing voxel-wise features and aggregating tumor subregions with similar radiological phenotypes (imaging habitats) to elucidate the distribution of tumor heterogeneity within and among tumors. Despite the promising applications of imaging habitats, they may be affected by variability of radiomics features, preventing comparison and generalization of imaging habitats techniques. We performed a comprehensive repeatability and reproducibility analysis of voxel-wise radiomics features in more than 500 lung cancer patients with computed tomography (CT) images and demonstrated the effect of voxel-wise radiomics variability on imaging habitats computation in 30 lung cancer patients with test–retest images. Repeatable voxel-wise features characterized texture heterogeneity and were reproducible regardless of the applied feature extraction parameters. Imaging habitats computed using robust radiomics features were more stable than those computed using all features in test–retest CTs from the same patient. Nine voxel-wise radiomics features (joint energy, joint entropy, sum entropy, maximum probability, difference entropy, Imc1, Imc2, Idn and Idmn) were repeatable and reproducible. This supports their application for computing imaging habitats in lung tumors towards the discovery of previously unseen tumor heterogeneity and the development of novel non-invasive imaging biomarkers for precision medicine.

IAtbP-Hyb-EnC: Prediction of antitubercular peptides via heterogeneous feature representation and genetic algorithm based ensemble learning model.

Tuberculosis (TB) is a worldwide illness caused by the bacteria Mycobacterium tuberculosis. Owing to the high prevalence of multidrug-resistant tuberculosis, numerous traditional strategies for developing novel alternative therapies have been presented. The effectiveness and dependability of these procedures are not always consistent. Peptide-based therapy has recently been regarded as a preferable alternative due to its excellent selectivity in targeting specific cells without affecting the normal cells. However, due to the rapid growth of the peptide samples, predicting TB accurately has become a challenging task. To effectively identify antitubercular peptides, an intelligent and reliable prediction model is indispensable. An ensemble learning approach was used in this study to improve expected results by compensating for the shortcomings of individual classification algorithms. Initially, three distinct representation approaches were used to formulate the training samples: k-space amino acid composition, composite physiochemical properties, and one-hot encoding. The feature vectors of the applied feature extraction methods are then combined to generate a heterogeneous vector. Finally, utilizing individual and heterogeneous vectors, five distinct nature classification models were used to evaluate prediction rates. In addition, a genetic algorithm-based ensemble model was used to improve the suggested model's prediction and training capabilities. Using Training and independent datasets, the proposed ensemble model achieved an accuracy of 94.47% and 92.68%, respectively. It was observed that our proposed "iAtbP-Hyb-EnC" model outperformed and reported ~10% highest training accuracy than existing predictors. The "iAtbP-Hyb-EnC" model is suggested to be a reliable tool for scientists and might play a valuable role in academic research and drug discovery. The source code and all datasets are publicly available at https://github.com/Farman335/iAtbP-Hyb-EnC.

Feature extraction and machine learning techniques for identifying historic urban environmental hazards: New methods to locate lost fossil fuel infrastructure in US cities.

U.S. cities contain unknown numbers of undocumented “manufactured gas” sites, legacies of an industry that dominated energy production during the late-19th and early-20th centuries. While many of these unidentified sites likely contain significant levels of highly toxic and biologically persistent contamination, locating them remains a significant challenge. We propose a new method to identify manufactured gas production, storage, and distribution infrastructure in bulk by applying feature extraction and machine learning techniques to digitized historic Sanborn fire insurance maps. Our approach, which relies on a two-part neural network to classify candidate map regions, increases the rate of site identification 20-fold compared to unaided visual coding.

Algorithmically Exploiting the Knowledge Accumulated in Textual Domains for Technical Support

Processing natural language and extract relevant information in deep technical engineering domain remains an open challenge. 
 On the other side, manufacturers of high-value assets which often deliver product services through the equipment life, supporting maintenance, spare parts management and remote monitoring and diagnostics for issues resolution, have availability of a good amount of textual data containing technical cases with a certain engineering depth.
 This paper presents a case study in which various Artificial Intelligence algorithms were applied to historical technical cases to extract know-how useful to help technicians in approaching new cases.
 Initially the work process and available data are presented; the focus is on the outbound communication delivered from the technical team to the site operators, that is structured in 3 main paragraphs: event description, technical assessment, recommended actions. 
 The work proceeded in two parallel streams: the first concerned the analysis of event descriptions and technical assessments, aiming to detect recurring topics; the second concerned the analysis of recommended actions that technical support delivered trough years to site operators in order to create a library, which can help for enabling statistical data analysis, quality check review and being the starting point for further AI/NLP developments.
 A text preprocessing was applied to both streams, consisted in defining standard and domain entities / stopwords and identifying / removing them, creating acronyms and synonyms maps in order to make context disambiguation, sentence splitting for the recommended actions, and finally text lemmatization. For every text the output of the preprocess was a series of keywords. 
 Then, unsupervised learning algorithms were applied. For this purpose, firstly, we applied feature extraction, bag of words (TF-IDF) and word embeddings (W2V, D2V, BERT), in order to transform our data from language domain into points in a n-features domain. Afterwards, different combinations of unsupervised algorithms were applied to split data into homogeneous groups, such as: LDA, K-means, Spectral, Affinity Propagation and HDBSCAN. 
 The combinations between language modeling and clustering were evaluated using the Silhouette score and visual analysis. 
 To validate the effectiveness, the developed NLP algorithms have been implemented into the current SW application used by technical support to perform the service. Moreover, a dedicated app to show trending topics and retrieve insightful information has been developed. 
 An outlook of the open technical challenges and on the future perspective of NLP applications in the work process are finally delivered.