A. Çakır

Named entity recognition (NER) is an extensively studied task that extracts and classifies named entities in a text. NER is crucial not only in downstream language processing applications such as relation extraction and question answering but also in large scale big data operations such as real-time analysis of online digital media content. Recent research efforts on Turkish, a less studied language with morphologically rich nature, have demonstrated the effectiveness of neural architectures on well-formed texts and yielded state-of-the art results by formulating the task as a sequence tagging problem. In this work, we empirically investigate the use of recent neural architectures (Bidirectional long short-term memory (BiLSTM) and Transformer-based networks) proposed for Turkish NER tagging in the same setting. Our results demonstrate that transformer-based networks which can model long-range context overcome the limitations of BiLSTM networks where different input features at the character, subword, and word levels are utilized. We also propose a transformer-based network with a conditional random field (CRF) layer that leads to the state-of-the-art result (95.95% f-measure) on a common dataset. Our study contributes to the literature that quantifies the impact of transfer learning on processing morphologically rich languages.

This paper presents an extensive empirical study on the integration of dimensionality reduction techniques with advanced unsupervised time series anomaly detection models, focusing on the MUTANT and Anomaly-Transformer models. The study involves a comprehensive evaluation across three different datasets: MSL, SMAP, and SWaT. Each dataset poses unique challenges, allowing for a robust assessment of the models' capabilities in varied contexts. The dimensionality reduction techniques examined include PCA, UMAP, Random Projection, and t-SNE, each offering distinct advantages in simplifying high-dimensional data. Our findings reveal that dimensionality reduction not only aids in reducing computational complexity but also significantly enhances anomaly detection performance in certain scenarios. Moreover, a remarkable reduction in training times was observed, with reductions by approximately 300\% and 650\% when dimensionality was halved and minimized to the lowest dimensions, respectively. This efficiency gain underscores the dual benefit of dimensionality reduction in both performance enhancement and operational efficiency. The MUTANT model exhibits notable adaptability, especially with UMAP reduction, while the Anomaly-Transformer demonstrates versatility across various reduction techniques. These insights provide a deeper understanding of the synergistic effects of dimensionality reduction and anomaly detection, contributing valuable perspectives to the field of time series analysis. The study underscores the importance of selecting appropriate dimensionality reduction strategies based on specific model requirements and dataset characteristics, paving the way for more efficient, accurate, and scalable solutions in anomaly detection.

If new phenomena beyond the Standard Model will be discovered at the LHC, the properties of the new particles could be determined with data from the High-Luminosity LHC and from a future linear collider like the ILC. We discuss the possible interplay between measurements at the two accelerators in a concrete example, namely a full SUSY model which features a small $$ \widetilde{\tau }_1$$ -LSP mass difference. Various channels have been studied using the Snowmass 2013 combined LHC detector implementation in the Delphes simulation package, as well as simulations of the ILD detector concept from the Technical Design Report. We investigate both the LHC and the ILC capabilities for discovery, separation and identification of various parts of the spectrum. While some parts would be discovered at the LHC, there is substantial room for further discoveries at the ILC. We finally highlight examples where the precise knowledge about the lower part of the mass spectrum which could be acquired at the ILC would enable a more in-depth analysis of the LHC data with respect to the heavier states.

Abstract Today we create and collect more data than we have in the past. All this data comes from different sources, including social media platforms, our phones and computers, healthcare gadgets and wearable technology, scientific instruments, financial institutions, the manufacturing industry, news channels, and more. When these data are analyzed in a real-time nature, it offers businesses the opportunity to take quick action in business-development processes (B2B, B2C), gain a different perspective, and better understand applications, creating new opportunities. While changing their sales and marketing strategies, businesses are now able to manage the data they collect in real-time to transform themselves, to record them in a healthy way, to analyze and evaluate data-based processes, and to determine their digital transformation roadmaps, their interactions with their customers, sectoral diffraction, application, and analysis. They want to accelerate the transformation processes within the technology triangle. Thus, big data, recently called as small and wide data, is at the center of everything and becomes an important application for digital transformation. Digital transformation helps companies embrace change and stay competitive in an increasingly digital world. The value of big data in manufacturing, independent from sectoral variations, comes from its ability to combine both in an organization’s efforts to both digitize and automate its end-to-end business operations. In this study, the current digitalization and automation applications of one of the plastic injection-based manufacturing companies at scale will be discussed. Presented open-source-based big data analytics platform, DataCone, that increases data processing efficiency, storage optimization, encourages innovation for real time monitorization and analytics, and support new business models in different industry segments will be demonstrated and discussed. Thus, development and applied ML solutions will be discussed providing important prospects for the future.

In this work, the simulation of optical photons is carried out in an antineutrino detector module consisting of a plastic scintillator connected to light guides and photomultipliers on both ends, which is considered to be used for remote reactor monitoring in the field of nuclear safety. Using Monte Carlo (MC) based GEANT4 simulation, numerous parameters influencing the light collection and thereby the energy resolution of the antineutrino detector module are studied: e.g., degrees of scintillator surface roughness, reflector type, and its applying method onto scintillator and light guide surface, the reflectivity of the reflector, light guide geometries and diameter of the photocathode. The impact of each parameter is investigated by looking at the detected spectrum, i.e. the number photoelectrons per depositing energy. In addition, the average light collection efficiency of the detector module and its spatial variation are calculated for each simulation setup. According to the simulation results, it is found that photocathode size, light guide shape, reflectivity of reflecting material and wrapping method show a significant impact on the light collection efficiency while scintillator surface polishing level and the choose of reflector type show relatively less impact. This study demonstrates that these parameters are very important in the design of plastic scintillator included antineutrino detectors to improve the energy resolution efficiency.

Digitization and data-driven manufacturing process is needed for today's industry. The term Industry 4.0 stands for today industrial digitization which is defined as a new level of organization and control over the entire value chain of the life cycle of products; it is geared towards increasingly individualized customer's high-quality expectations. However, due to the increase in the number of connected devices and the variety of data, it has become difficult to store and analyze data with conventional systems. The motivation of this paper is to provide an overview of the understanding of the big data pipeline, providing a real-time on-premise data acquisition, data compression, data storage and processing with Apache Kafka and Apache Spark implementation on Apache Ha-doop cluster, and identifying the challenges and issues occurring with implementation the Farplas manufacturing company, which is one of the biggest Tier 1 automotive supplier in Turkey, to study the new trends and streams related to topics via Industry 4.0.

The search for new physics is a major goal of the LHC physics program. As excitement grows for the upcoming start of Run 2, I review the CMS and ATLAS searches for physics beyond the Standard Model from Run 1 and present recent analyses. These searches have covered a wide range of new physics scenarios including Supersymmetry, new resonances, additional Higgs bosons, new hidden sectors, other Dark Matter, and multi-charged particles. In addition to reviewing some of the techniques that made the analyses possible, I will summarize what we have learned from the results and briefly discuss prospects for Run 2.

Particles produced by nuclear decay, cosmic radiation and reactions can be identified through various methods. One of these methods that has been effective in the last century is the cloud chamber. The chamber makes visible cosmic particles that we are exposed to radiation per second. Diffusion cloud chamber is a kind of cloud chamber that is cooled by dry ice. This traditional model has some application difficulties. In this work, Peltier-based cloud chamber cooled by thermoelectric modules is studied. The new model provided uniformly cooled base of the chamber, moreover, it has longer lifetime than the traditional chamber in terms of observation time. This gain has reduced the costs which spent each time for cosmic particle observation. The chamber is an easy-to-use system according to traditional diffusion cloud chamber. The new model is portable, easier to make, and can be used in the nuclear physics experiments. In addition, it would be very useful to observe Muons which are the direct evidence for Lorentz contraction and time expansion predicted by Einsteins special relativity principle.

We compare existing segmented plastic antineutrino detectors with our new geometrically improved design for antineutrino detection and light collection efficiency. The purpose of this study is to determine the most suitable design style for remote reactor monitoring in the context of nuclear safeguards. Using Monte Carlo based GEANT4 simulation package, we perform detector simulation based on two prominent experiments: Plastic antineutrino detector array (Panda) and Core monitoring by reactor antineutrino detector (Cormorad). In addition to these two well-known designs, another concept, the Panda2, can be obtained by making a small variation of Panda detector, is also considered in the simulation. The results show that the light collection efficiency of the Cormorad is substantially less with respect to the other two detectors while the highest antineutrino detection efficiency is achieved with the Cormorad and Panda2. Furthermore, as an alternative to these design choices, which are composed of an array of identical rectangular-shaped modules, we propose to combine regular hexagonal-shaped modules which minimizes the surface area of the whole detector and consequently reduces the number of optical readout channels considerably. With this approach, it is possible to obtain a detector configuration with a slightly higher detection efficiency with respect to the Panda design and a better energy resolution detector compared to the Cormorad design.

This study presents a new concept of segmented antineutrino detector based on hexagonal plastic scintillator bars for detecting antineutrinos from a nuclear reactor core. The choice of hexagonal scintillator bars is original and provides compactness. The proposed detector detects antineutrinos via inverse beta decay (IBD) with the prompt-delayed double coincidence. Owing to its segmented structure, the background, which satisfies the delayed coincidence condition can be eliminated by applying proper event selection cuts. In this manner, the main focus is to determine proper selection criteria to precisely tag the true IBD events. Monte-Carlo simulation is carried out to understand the characteristic of the IBD interaction in the proposed detector by using Geant4 toolkit. A set of event selection criteria is established based on the simulated data. It is found that a detection efficiency of 10% can be achieved with the selection condition applied. It is also shown that fast neutrons, which constitute the main background source for above-ground detection, can be effectively eliminated with these selection criteria. The motivation for this study is to install this compact detector at a short distance (<100 m) from the Akkuyu Nuclear Power Plant, which is expected to start operation in 2023.

This work addresses an alternative approach for query expansion (QE) using a generative adversarial network (GAN) to enhance the effectiveness of information search in e-commerce. We propose a modified QE conditional GAN (mQE-CGAN) framework, which resolves keywords by expanding the query with a synthetically generated query that proposes semantic information from text input. We train a sequence-to-sequence transformer model as the generator to produce keywords and use a recurrent neural network model as the discriminator to classify an adversarial output with the generator. With the modified CGAN framework, various forms of semantic insights gathered from the query document corpus are introduced to the generation process. We leverage these insights as conditions for the generator model and discuss their effectiveness for the query expansion task. Our experiments demonstrate that the utilization of condition structures within the mQE-CGAN framework can increase the semantic similarity between generated sequences and reference documents up to nearly 10% compared to baseline models

This work addresses an alternative approach for query expansion (QE) using a generative adversarial network (GAN) to enhance the effectiveness of information search in e-commerce. We propose a modified QE conditional GAN (mQE-CGAN) framework, which resolves keywords by expanding the query with a synthetically generated query that proposes semantic information from text input. We train a sequence-to-sequence transformer model as the generator to produce keywords and use a recurrent neural network model as the discriminator to classify an adversarial output with the generator. With the modified CGAN framework, various forms of semantic insights gathered from the query-document corpus are introduced to the generation process. We leverage these insights as conditions for the generator model and discuss their effectiveness for the query expansion task. Our experiments demonstrate that the utilization of condition structures within the mQE-CGAN framework can increase the semantic similarity between generated sequences and reference documents up to nearly 10\% compared to baseline models.

GİRİŞBir kişinin geleceği üzerinde en fazla etkiye sahip olan kurum hiç şüphesiz okuludur

The proliferation of both internet usage and users have been remarkably increased due to certain situations that influenced face-to-face communications, which in turn have created high pressure on Internet Service Providers (ISPs). This research mainly aims to boost ISP services by conducting near real-time analysis for customer’s behavior movements based on their score of central Points of Presence (POP). In addition, this study focuses on establishing special Recurrent Artificial Intelligence (RNN) architecture to make daily sales predictions based on various central POPs. The process utilizes different RNN architectures, Long Short Time Memory (LSTM) and Gated Recurrent Unit (GRU), and compares them in order to make smart scoring measurements for customers’ high-dimensional data. As a result, it can be concluded that LSTM architecture has achieved much better Mean squared Error (MSE) than GRU architecture. LSTM outperforms GRU in forecasting less sensitive outliers, with an average Mean Absolute Error (MAE) of 1.354 for LSTM and 1.554 for GRU. Additionally, LSTM performs better in forecasting outliers, with an average MSE of 3.592 compared to GRU’s average of 4.8. Thereafter, the obtained results are merged over private Application Programming Interface (API) and monitored over smart reports. Eventually, the outcomes of this research can be summarized in providing several benefits for customers such as increasing internet performance, reaching promised speed, and shortening activation times. ISP-related benefits such as gaining reputation, promoting sales, and reducing customers’ negative support tickets can be achieved as well.

In this study, we propose an innovative approach to query expansion (QE) in e-commerce, aiming to enhance the effectiveness of information search. Our method utilizes a generative adversarial network (GAN) called modified QE conditional GAN (mQE-CGAN) to expand queries by generating synthetic queries that incorporate semantic information from textual input. The (mQE-CGAN) framework consists of a generator and a discriminator. The generator is a sequence-to-sequence transformer model trained to produce keywords, while the discriminator is a recurrent neural network model used to classify the generator’s output in an adversarial manner. By incorporating a modified CGAN framework, we introduce various forms of semantic insights from the query-document corpus into the generation process. These insights serve as conditions for the generator model and are instrumental in improving the query expansion task. Through various preliminary experiments, we demonstrate that the utilization of condition structures within the mQE-CGAN framework significantly enhances the semantic similarity between the generated sequences and reference documents. Compared to baseline models, our approach achieves an impressive increase of approximately 5–10% in semantic similarity.

The purpose of this study is to determine whether the conflict management strategies used by academic staff influence organizational commitment and whether they differ according to independent variables.In this study, a relational screening model, one of the research methods, was used.The accessible population of the research consists of 106 academic staff working in Vocational Schools of Aydın Adnan Menderes University.In the research conducted in the 2018-2019 academic year, Conflict Action Styles Inventory and Organizational Commitment Scale were used as data collection tools.Statistical analyses were performed with IBM SPSS 25.0 program.Descriptive statistics were expressed as frequency (f), percentage (%), mean (X), standard deviation (SD), minimum and maximum values.In the study, firstly, whether the scales fit the normal distribution hypothesis was determined by skewness and kurtosis coefficients and parametric test methods were preferred.The gender, age, marital status, education, academic title, and term of office distribution of the academic staff included in the study were examined and it was observed with the regression model created in line with the determined purpose conflict action styles predict organizational commitment positively and statistically significant.Moreover, when other variables are held constant, a one-unit increase in the level of conflict action styles of academic staff provides an increase of 0.183 in the level of organizational commitment.Conflict action styles of academic staff should be studied with different groups at the point of organizational commitment, opportunities should be given for employees to improve themselves and rise in other institutions, and the importance of creating organizational commitment should be known by taking opinions not only from academicians but also from other professions, and studies related to conflict styles and organizational commitment should be conducted at various times.

This work addresses an alternative approach for query expansion (QE) using a generative adversarial network (GAN) to enhance the effectiveness of information search in e-commerce. We propose a modified QE conditional GAN (mQE-CGAN) framework, which resolves keywords by expanding the query with a synthetically generated query that proposes semantic information from text input. we train a sequence-to-sequence transformer model as the generator to produce keywords and use a recurrent neural network model as the discriminator to classify an adversarial output with the generator. with the modified CGAN framework, Various forms of semantic insights gathered from the query-document corpus are introduced to the generation process. we leverage these insights as conditions for the generator model and discuss their effectiveness for the query expansion task. our experiments demonstrate that the utilization of condition structures within the mQE-CGAN framework can increase the semantic similarity between generated sequences and reference documents up to nearly 10% compared to baseline models.

Robotic arms are widely prevalent and find utility in a variety of applications. However, a significant and widespread challenge faced by these arms is their inability to replicate the intricate functionalities of a human hand, primarily due to the distinct structure of the hand. The primary objective of this work is to create a simulation of a robotic arm that can replicate the movements and functions of a human hand in real-time. Data obtained from hand and arm gestures created with Mediapipe will be transferred in real-time to the robotic arm that is visualized and simulated on ROS and Gazebo. Thus, the hand and arm movements of the user in front of the camera will be effective in real-time on the manipulable joints. This advancement holds the potential to facilitate the construction of a robot capable of emulating both the hand and the arm of humans with high fidelity, thereby enabling comprehensive control over the robotic arm’s actions in real-time.

Abstract. One of the common problems at the intersection of geographical information science and transportation science is the estimation of origin-destination (OD) matrices. The emergence of sensor technologies offers unprecedented opportunities in this regard since massive amounts of traffic data can be collected in an easy way. Researchers and practitioners need to choose a suitable DataBase Management System (DBMS) among alternatives, such that storing and analysing traffic data to estimate the OD matrix is feasible. The aim of this paper is to compare the performance of two such notable DBMSs, PostgreSQL and MongoDB, in the context of OD matrix estimation. The experiments are carried out on New York City’s openly available taxi data on two different polygon sets: taxi zones and census blocks. These polygon layers consist of 263 and 38794 features respectively. The results suggest that Postgres outperforms MongoDB by generating the OD matrix instantly. The run time of MongoDB varies depending on the analysed time interval and follows a trip demand curve. As there are more trips involved in the generation of the OD matrix, so does the execution time increases in MongoDB. On the other hand, the query results are the same. Finally, the origin points of the taxi trips are visualised in QGIS using the ‘TimeManager’ plugin, and results are presented through a web-interface.

The Coast Guard Command, which has a wide range of duties as saving human lives, protecting natural resources, preventing marine pollution and battle against smuggling, uses diesel main engines in its ships, as in other military and commercial ships. It is critical that the main engines operate smoothly at all times so that they can respond quickly while performing their duties, thus enabling fast and early detection of faults and preventing failures that are costly or take longer to repair. The aim of this study is to create and to develop a model based on current data, to select machine learning algorithms and ensemble methods, to develop and explain the most appropriate model for fast and accurate detection of malfunctions that may occur in 4-stroke high-speed diesel engines. Thus, it is aimed to be an exemplary study for a data-based decision support mechanism.

We create and collect more data today than we have in the past. All this data comes from and is reviewed from different sources, including social media platforms, our phones and computers, healthcare gadgets and wearable technology, scientific instruments, financial institutions, the manufacturing industry, news channels and more. When these small and wide data are analyzed, it offers businesses the opportunity to take quick action in business-development processes (B2B, B2C), gain a different perspective, and better understand applications, creating new opportunities. While changing their sales and marketing strategies, businesses are now able to manage the data they collect in real time to transform themselves, to record them in a healthy way, to analyze and evaluate data-based processes, and to determine their digital transformation roadmaps, their interactions with their customers, sectoral diffraction, application and advanced analyses. They want to accelerate the transformation processes within the technology triangle. As a result, big data technologies and applications are at the center of everything and becomes an important application for digital transformation.

The search for new physics is a major goal of the LHC physics program. As excitement grows for the upcoming start of Run 2, I review the CMS and ATLAS searches for physics beyond the Standard Model from Run 1 and present recent analyses. These searches have covered a wide range of new physics scenarios including Supersymmetry, new resonances, additional Higgs bosons, new hidden sectors, other Dark Matter, and multi-charged particles. In addition to reviewing some of the techniques that made the analyses possible, I will summarize what we have learned from the results and briefly discuss prospects for Run 2.

Integration of superpartners out of the spectrum induces potentially large contributions to Yukawa couplings. Supersymmetric threshold corrections therefore influence the CKM matrix prediction in a non-trivial way. We study the effects of threshold corrections on high-scale flavor structures specified at the gauge coupling unification scale in supersymmetry. We first consider high-scale Yukawa textures which qualify as phenomenologically viable at tree level, and find that they are disqualified after incorporating the threshold corrections. Next, we consider Yukawa couplings, such as those with five texture zeroes, which are incapable of explaining flavor-changing processes. Incorporating threshold corrections, however, makes them phenomenologically viable textures. Therefore, supersymmetric threshold corrections are found to have an observable impact on Yukawa couplings of quarks, and any confrontation of high-scale textures with experiments at the weak scale must take into account such corrections. PACS Codes: 12.60.Jv, 12.15.Hh

Text analysis in the processing of deriving information from social media channels has become more crucial in recent years. These studies can be applied to social media data to answer wide variety of questions about consumers, brands, products or any other campaign strategies for the content producers. In this context, understand the sentiment or specific emotions expressed during the social media communication channels, mostly on Twitter, identify to intent about the corresponding content and any other stages of the consumer interest. In this study, one of the biggest Turkish news provider, Hürriyet, and its top-rated authors’ tweets are analyzed by using advanced natural language processing techniques. Due to the limitation of supported libraries for the Turkish language, preprocessing steps for unstructured datasets are studied with self-developed classes using machine learning techniques. Combining Twitter circulation together with the company’s clickstream data, we motivate to find the unique patterns for user’s attention on specific categories on the news content. By measuring the full value of digital media attention data through Twitter and click datasets, we will provide the ideal scenario for using controlled marketing experiments at the digital media sector in Turkey.

The readers’ behavior have partially turned into consuming news online, after news providers are transformed the printed newspapers into digital web sites. Engaging attention of the readers by retaining them on the web site contents as much as possible to consume more articles has become crucial and as a scarce resource, it is also challenging for the providers. While readers click through on the pages, system records each click for classical counting approach, also measure the time spent on pages which is critical for news providers in digital web sites. In this regard, recommendation engines which builds on the natural language processing with machine learning algorithms play important role to gain insights from the vectorized contents and provide similar articles to the readers. We motivate to analyze news content similarities expanding with tags and emotion similarities by comparing various text analytics methods results starting from sports and travel categories. By adding domain knowledge on the findings, our aim is to have a specific category based approach which could be applied on different news providers contents in Turkey.

After the observation of a Higgs boson near 125 GeV, the high energy physics community is investigating possible next steps for entering into a new era in particle physics. It is planned that the Large Hadron Collider will deliver an integrated luminosity of up to 3000/fb for the CMS and ATLAS experiments, requiring several upgrades for all detectors. The reach of various representative searches for supersymmetry and exotica physics with the upgraded detectors are discussed in this context, where a very high instantaneous luminosity will lead to a large number of pileup events in each bunch crossing. This note presents example benchmark studies for new physics prospects with the upgraded ATLAS and CMS detectors at a centre-of-mass energy of 14 TeV. Results are shown for an integrated luminosity of 300/fb and 3000/fb.

The search for new physics is a major goal of the LHC physics program. As excitement grows for the upcoming start of Run 2, I review the CMS and ATLAS searches for physics beyond the Standard Model from Run 1 and present recent analyses. These searches have covered a wide range of new physics scenarios including Supersymmetry, new resonances, additional Higgs bosons, new hidden sectors, other Dark Matter, and multi-charged particles. In addition to reviewing some of the techniques that made the analyses possible, I will summarize what we have learned from the results and briefly discuss prospects for Run 2.

Various channels have been studied using the Snowmass 2013 combined LHC detector implementation in the Delphes simulation package, as well as simulations of the ILD detector concept from the Technical Design Report. We investigate both the LHC and ILC capabilities for discovery, separation and identication of various parts of the spectrum. While some parts would be discovered at the LHC, there is substantial room for further discoveries at the ILC. We nally highlight examples where the precise knowledge about the lower part of the mass spectrum which could be acquired at the ILC would enable a more in-depth analysis of the LHC data with respect to the heavier states.

The latest results from CMS on R-Parity violating Supersymmetry based on the 19.5/fb full dataset from the 8 TeV LHC run of 2012 are reviewed.The results are interpreted in the context of simplified models with multilepton and b-quark jets signatures that have low missing transverse energy arising from light top-squark pair with R-parity-violating decays of the lightest supersymmetric particle.In addition to simplified model, a new approach for phenomenological MSSM interpretation is shown which demonstrates that the obtained results from multilepton final states are valid for a wide range of supersymmetry models.

After the observation of a Higgs boson near 125 GeV, the high energy physics community is investigating possible next steps for entering into a new era in particle physics. It is planned that the Large Hadron Collider will deliver an integrated luminosity of up to 3000/fb for the CMS and ATLAS experiments, requiring several upgrades for all detectors. The reach of various representative searches for supersymmetry and exotica physics with the upgraded detectors are discussed in this context, where a very high instantaneous luminosity will lead to a large number of pileup events in each bunch crossing. This note presents example benchmark studies for new physics prospects with the upgraded ATLAS and CMS detectors at a centre-of-mass energy of 14 TeV. Results are shown for an integrated luminosity of 300/fb and 3000/fb.

After a very successful startup of the LHC in 2010, the CMS experiment has already accumulated significantly more data in 2011. After the successful re-discovery of the Standard Model, the search for signs of new physics has already reached, and in most cases enlarged, the limits from previous experiments. In this conference report I review the recent discovery reach of SUSY searches that will be performed with the 2011 data.

The latest results from CMS on R-Parity violating Supersymmetry based on the 19.5/fb full dataset from the 8 TeV LHC run of 2012 are reviewed. The results are interpreted in the context of simplified models with multilepton and b-quark jets signatures that have low missing transverse energy arising from light top-squark pair with R-parity-violating decays of the lightest supersymmetric particle. In addition to simplified model, a new approach for phenomenological MSSM interpretation is shown which demonstrates that the obtained results from multilepton final states are valid for a wide range of supersymmetry models.

Results of searches for SUSY production at CMS in events with third-generation signatures are presented. Along with missing energy, the final states may include hadronic jets with or without b-quark tag, light leptons, and tau leptons. These features serve both to distinguish standard-model components, and sensitivity to those SUSY models that lead to final states rich of heavy-flavored particles.

Ce memoire traite d’une problematique issue d’un besoin d’optimiser la performance de la chaine logistique et de reduction des stocks pour mieux ameliorer le stockage des produits finis. De meme etre une source de gains de productivite et d’efficacite dans une societe de l’industrie automobile, dont le but est de livrer aux clients en temps et en heure des produits de qualite au meilleur prix. A la suite de la delocalisation de certaines machines vient se greffer un nouveau projet la reorganisation des circuits de flux (reimplantation machines presses, zones de stockage...). Au travers de cette etude, il convient d’analyser comment les flux sont organises, reperer les dysfonctionnements et de determiner l’ensemble des stocks de piece de rechange reparti sur les 3 magasins de stockage (1 magasin de stockage sur site, 2 magasins de stockage externe). Les objectifs finaux sont d’avoir une meilleure organisation des zones logistiques, de rendre ceux-ci plus optimale et de reduire les couts de stockage et de transport (favoriser un gain de place au magasin le plus proche pour le stock de securite des produits finis).

After a very successful startup of the LHC in 2010, the CMS experiment has already accumulated significantly more data in 2011. After the successful re-discovery of the Standard Model, the search for signs of new physics has already reached, and in most cases enlarged, the limits from previous experiments. In this conference report I review the recent discovery reach of SUSY searches that will be performed with the 2011 data.

The latest results from CMS on R-Parity violating Supersymmetry based on the 19.5/fb full dataset from the 8 TeV LHC run of 2012 are reviewed. The results are interpreted in the context of simplified models with multilepton and b-quark jets signatures that have low missing transverse energy arising from light top-squark pair with R-parity-violating decays of the lightest supersymmetric particle. In addition to simplified model, a new approach for phenomenological MSSM interpretation is shown which demonstrates that the obtained results from multilepton final states are valid for a wide range of supersymmetry models.

After a very successful startup of the LHC in 2010, the CMS experiment has already accumulated significantly more data in 2011. After the successful re-discovery of the Standard Model, the search for signs of new physics has already reached, and in most cases enlarged, the limits from previous experiments. In this conference report I review the recent discovery reach of SUSY searches that will be performed with the 2011 data.

Results of searches for SUSY production at CMS in events with third-generation signatures are presented. Along with missing energy, the final states may include hadronic jets with or without b-quark tag, light leptons, and tau leptons. These features serve both to distinguish standard-model components, and sensitivity to those SUSY models that lead to final states rich of heavy-flavored particles.

Simplified models have become a widely used and important tool to cover the more diverse phenomenology beyond constrained SUSY models. However, they come with a substantial number of caveats themselves, and great care needs to be taken when drawing conclusions from limits based on the simplified approach. To illustrate this issue with a concrete example, we examine the applicability of simplified model results to a series of full SUSY model points which all feature a small stau-LSP mass difference, and are compatible with electroweak and flavor precision observables as well as current LHC results. Various channels have been studied using the Snowmass Combined LHC detector implementation in the Delphes simulation package, as well as the Letter of Intent or Technical Design Report simulations of the ILD detector concept at the ILC. We investigated both the LHC and ILC capabilities for discovery, separation and identification of all parts of the spectrum. While parts of the spectrum would be discovered at the LHC, there is substantial room for further discoveries and property determination at the ILC.

Precise determination of top-quark pairs is an essential tool for understanding the overall consistency of the standard model (SM) expectations, understanding limited New Physics (NP) models, through spin-spin correlation and polarization parameters, and has a critical impact on the analyses strategies at upcoming LHC programs. In this work, we review and discuss various state-of-the-art Monte Carlo (MC) methodologies as \textsc{MadGraph5}\_aMC@NLO, \textsc{Sherpa}, \textsc{Powheg-Box} and \textsc{Pythia8}, which are Matrix Element (ME)$/$Parton Shower (PS) matching generators including a complete set of spin correlation and polarization in top quark pair production with dileptonic final states. This is the first such study that not only compares the effects of different MC event generator approaches on spin density matrix elements and polarization parameters, but also investigates the effects of leading order (LO) and next-to-leading order (NLO) accuracy in QCD, and electroweak (EW) corrections via \textsc{Sherpa}. Moreover, as a continuation of the work, the prospects for possible NP scenarios through top-quark spin-spin correlation and polarization measurements for Supersymmetry (R parity conserved and violated models) and Dark Matter (top quarks associated mediator) models during upcoming LHC runs are briefly outlined. We find that all SM MC predictions for the defined set of variables are generally consistent with the experimental data and theoretical predictions within the uncertainty variations. Besides, for the distributions of the $\cos\varphi$, laboratory-frame observables ($\cos\varphi_{lab}$ and $|\Delta\phi_{ll}|$) and the observables generated by parity (P) and charge-parity (CP) conserving interactions, we conclude some clues that the considered signals and beyond may well be separated from experimental data and located above the SM predictions.

Clickstream data, which come with a massive volume generated by human activities on websites, have become a prominent feature for identifying readers' characteristics by newsrooms after the digitization of news outlets. Although the nature of clickstream data has a similar logic within websites, it has inherent limitations in recognizing human behaviours when looking from a broad perspective, which brings the need to limit the problem in niche areas. This study investigates the anonymized readers' click activities on the organizations' websites to identify news consumption patterns following referrals from Twitter,who incidentally reach but propensity is mainly routed news content. Methodologies for ensemble cluster analysis with mixed-type embedding strategies are applied and compared to find similar reader groups and interests independent of time. Various internal validation perspectives are used to determine the optimality of the quality of clusters, where the Calinski Harabasz Index (CHI) is found to give a generalizable result. Our findings demonstrate that clustering a mixed-type dataset approaches the optimal internal validation scores, which we define to discriminate the clusters and algorithms considering applied strategies when embedded by Uniform Manifold Approximation and Projection (UMAP) and using a consensus function as a key to access the most applicable hyperparameter configurations in the given ensemble rather than using consensus function results directly. Evaluation of the resulting clusters highlights specific clusters repeatedly present in the separated monthly samples by Adjusted Mutual Information scores greater than 0.5, which provide insights to the news organizations and overcome the degradation of the modeling behaviours due to the change in the interest over time.

Industry 4.0 is a motivation that represents the transformation by data-driven industrial operations and decision making by digitization of manufacturing processes to gain operational advantages in the market. Considering how the manufacturing sector is adopting data-driven operations is challenging, given that there is not a straightforward definition of machine traceability, receiving and storing raw data from manufacturing lines, gives an opportunity to analyse the processes in real time nature. Thanks to big data management platforms and artificial intelligence decision support algorithms, it gives the ability to deeply understand the complexity of the processes and, accordingly, to eliminate or minimise false methods and reduce the costs that are insufficient for production. In addition, one of the biggest preventable costs for metal machining processes is the tool breakage and tool wearing problems. The motivation of this paper is to discuss data-driven decision making possibilities of the tool wearing and optimise breakage costs with using artificial intelligence. Furthermore, the analysis provides a proof-of-concept that the existence of a digital infrastructure combined with the analytical capabilities, such as real-time data management and monitoring, and having a highly accurate LSTM based time-series integrated artificial intelligent predictive model, to deal with inefficiencies in production processes. To this end, in this context, by developing the latest advancements in big data analytics, we propose a scalable predictive and preventive maintenance architecture for metal machining processes domain. We also show the opportunities and challenges of utilizing the big data architecture in the manufacturing domain.

Milli Teknoloji Hamlesi, Türkiye’nin teknolojik ve ekonomik bağımsızlığını temin etmek adına tanımlanmış ülkemizin ihtiyaç duyduğu teknoloji ve yenilikçilikte rekabet edebilmesinin bir adımı olarak son on beş yılda ortaya konulmuştur. Milli Teknoloji Hamlesi’nin yapı taşları beş temel bileşenden oluşmakta ve yüksek eğitim kurumlarımız ile birlikte “Yüksek Teknoloji ve İnovasyon”, “Dijital Dönüşüm ve Sanayi Hamlesi”, “Girişimcilik”, “Beşerî Sermaye” ve “Altyapı” esaslarına göre ihtiyaç duyulan rekabetçiliği canlandırmak üzere yürütülmektedir. Etkin ve verimli öncül programlar kapsamında yerinde Ar-Ge, konu odaklı geliştirme ve ürünleştirme çalışmaları ile yeni teknolojilerin üretimi konusunda özel sektöre öncülük eden ve koordineli çalışan bir yapı kurulmuştur. Milli Teknoloji Hamlesi âdeta bir milli mücadele olarak tanımlanmakta, ülkemizi teknoloji ve sanayi alanında küresel bir aktör haline getirmek için başta araştırma üniversiteleri olmak üzere kapsamlı bir çalışma yürütülmektedir. Kritik teknolojileri yerli ve milli olarak geliştirmek, yüksek teknoloji alanlarında rekabetçi araştırma, geliştirme, ürün ve hizmetler sunmak, özgün ve yenilikçi üretimle küresel değer zincirlerindeki payımızı artırmak Milli Teknoloji Hamlesi’nin ana hedefleri arasında yer almaktadır. Yetişmiş insan gücünün ve altyapı kullanımlarının artırılması, girişim ekosisteminin genişletilmesi, disiplinlerarası katılımcı paydaşların bu ekosistemdeki varlıkları teknolojinin ve refahın bağımsız hale getirilmesi için uzun soluklu bir seferberlik sürecinin başlangıcı durumundadır. Türkiye’nin uluslararası alanda öncü olabilecek Ar-Ge altyapısı, yetişmiş insan gücü, yenilikçi iş modelleri, ürünler ve hizmetler çıkarabilmesi güçlü bir yükseköğretim ve girişimcilik ekosistemine sahip olmasına bağlıdır. Araştırmacılarımız ve girişimcilerimizin küresel gelişimi iyi okuyabilmeleri Türkiye’nin uluslararası pazarlara açılmasını sağlayacak ve böylece Türkiye gelecekte daha güçlü bir yapıya güçlü üniversiteleri aracılığı ile ulaşacaktır.

The National Technology Initiative has been put forward in the last fifteen years as a step towards the technology and innovation competitiveness needed by our country, which has been defined in order to ensure Türkiye’s economic and technological independence. The building blocks of the National Technology Initiative consist of five basic components and are carried out together with our higher education institutions to revive the competitiveness needed according to the principles of “High Technology and Innovation”, “Digital Transformation and Industry Initiative”, “Entrepreneurship”, “Human Capital” and “Infrastructure”. Within the scope of an effective and efficient program, a structure that leads the private sector and works in coordination with on-site R&amp;D, subject-oriented development and productization studies and the production of new technologies has been established. The National Technology Initiative is defined as a national struggle, and a comprehensive study is carried out, especially in research universities, in order to make our country a global actor in the field of technology and industry. Developing critical technologies locally and nationally, providing competitive research, development, products and services in high-tech fields, increasing our share in global value chains with original and innovative production are among the main objectives of the National Technology Initiative. It is the beginning of a long-term mobilization process in order to increase the use of trained manpower and infrastructure, expand the entrepreneurial ecosystem, and make the existence of interdisciplinary participatory stakeholders in this ecosystem independent of technology and welfare. The fact that Türkiye can be a pioneer in the international arena, R&amp;D infrastructure, trained manpower, innovative business models, products and services depends on having a strong higher education and entrepreneurship ecosystem. In this manner, thanks to our researchers and entrepreneurs, who can read the global economy well and will enable Türkiye to open up to international markets. Türkiye will achieve a stronger structure in the future through its strong universities, with the initiative of national technology.

Plasma Wakefield Accelerators promise huge acceleration gradients that are three orders of magnitude greater than today's conventional radio frequency (RF) accelerators. These novel accelerators show also the potential of diminishing the size of the future accelerators nearly by the same factor. This review gives brief explanations and the working principles of the Plasma Wakefield Accelerators and shows the recent developments of the field. The current challenges are given and the potential future use of the Plasma Wakefield Accelerators are discussed.

In this study 136 water samples used as drinking water in Balıkesir / Erdek, Mersin / Karaduvar-Davultepe, Samsun / Merkez and Diyarbakır / Central regions located in different regions of Anatolia were evaluated.In addition to their drinkability characteristics, chemical and physical parameters were determined.Element and anion contents of drinking water samples taken from the regions determined by ICP-MS (B, Na, Mg, K, Ca, Cr, Mn, Ni, Cu, Zn, As, Cd, Ba, Pb, Al, Fe, Se, Mo) and by the ion chromatography (F -, Cl -, NO2 -, NO3 -, Br -, SO4 -2 , PO4 -2 ).Analysis results were evaluated in compared with TSE 266 standard, WHO (World Health Organization) and EPA (USD Environmental Protection Agency); Balıkesir/Erdek (B: 0.13, Al: 2.08, Fe: 0.44, Mn: 0.10, Cl -:287.06,NO2 -:23.54,NO3:179.41 -, Br -:0.64,SO4 -2 :302.61ppm), Mersin/Karaduvar-Davultepe (Mg: 130.30, Al: 0.24, As: 0.

Görüşümüzce, bütünleşmede dışsal değişkenlere ilişkin bir çalışmanın öncelikle Neumann’ınki türünden bir süreklilik üzerindeki yerini saptaması gerekir (bu, yaklaşımda ne ölçüde içsel ne ölçüde de dışsal değişkenlere yer verileceğinin kararlaştırılması demektir). Ardından, içsel ve dışsal değişkenlerin geçerliğini güvenceye almak için Waever 'in Avrupa tanımlarından birine ya da görece güncelleştirilmiş diğer bir Avrupa tanımına gönderme yoluyla Avrupa'nın sınırlan kararlaştırılmalıdır. Dışsal değişkenler alanına girilmesinin ardından bu alanda bir bölütleme gerçekleştirilmelidir. Şimdiki durumda hazır bir bölütleme kalıbı olmadığından kuramcı kendince bir bölütlemeye gitmekte özgür gözükmektedir. Bu aşamada, yazımız, bir seçenek olmak üzere yalın bir bölütleme kalıbı önermektedir: dışsal değişkenler ilkin, işletsel bir benzetmeyle, ‘çeken etkenler’ ve 'iten etkenler’ olarak ikiye ayrılmaktadır. Ardından iten etkenler de kendi içinde ‘bölücü’ ve ‘bütünleştirici ’ iten etkenler diye yeniden bölütlenmektedir.

Named entity recognition (NER) is an extensively studied task that extracts and classifies named entities in a text. NER is crucial not only in downstream language processing applications such as relation extraction and question answering but also in large scale big data operations such as real-time analysis of online digital media content. Recent research efforts on Turkish, a less studied language with morphologically rich nature, have demonstrated the effectiveness of neural architectures on well-formed texts and yielded state-of-the art results by formulating the task as a sequence tagging problem. In this work, we empirically investigate the use of recent neural architectures (Bidirectional long short-term memory and Transformer-based networks) proposed for Turkish NER tagging in the same setting. Our results demonstrate that transformer-based networks which can model long-range context overcome the limitations of BiLSTM networks where different input features at the character, subword, and word levels are utilized. We also propose a transformer-based network with a conditional random field (CRF) layer that leads to the state-of-the-art result (95.95\% f-measure) on a common dataset. Our study contributes to the literature that quantifies the impact of transfer learning on processing morphologically rich languages.

Domain Name System (DNS) is a hierarchical distributed naming system for computers, services, or any resource connected to the Internet. A DNS resolves queries for URLs into IP addresses for the purpose of locating computer services and devices worldwide. As of now, analytical applications with a vast amount of DNS data are a challenging problem. Clustering the features of domain traffic from a DNS data has given necessity to the need for more sophisticated analytics platforms and tools because of the sensitivity of the data characterization. In this study, a cloud based big data application, based on Apache Spark, on DNS data is proposed, as well as a periodic trend pattern based on traffic to partition numerous domain names and region into separate groups by the characteristics of their query traffic time series. Preliminary experimental results on a Turknet DNS data in daily operations are discussed with business intelligence applications.

The gap between the marketer and the customer is increasing in the recent years. The marketers are not able to accurately segment customers. Predictive modelling and auto-optimization technologies will be disrupting the digital customer experience delivery space. Being able to predict the future behavior of online and mobile visitors, the gap between marketers and customers will decrease. In this study, with the clickstream data that have been collected from the users on the websites, machine learning models will be created to predict each and every users’ likelihood to purchase, so that the marketers can target only those users, in order to have higher ROI’s in advertising world.

Domain Name System (DNS) is a hierarchical distributed naming system for computers, services, or any resource connected to the Internet. A DNS resolves queries for URLs into IP addresses for the purpose of locating computer services and devices worldwide. As of now, analytical applications with a vast amount of DNS data are a challenging problem. Clustering the features of domain traffic from a DNS data has given necessity to the need for more sophisticated analytics platforms and tools because of the sensitivity of the data characterization. In this study, a cloud based big data application, based on Apache Spark, on DNS data is proposed, as well as a periodic trend pattern based on traffic to partition numerous domain names and region into separate groups by the characteristics of their query traffic time series. Preliminary experimental results on a Turknet DNS data in daily operations are discussed with business intelligence applications.

Named entity recognition (NER) is an extensively studied task that extracts and classifies named entities in a text. NER is crucial not only in downstream language processing applications such as relation extraction and question answering but also in large scale big data operations such as real-time analysis of online digital media content. Recent research efforts on Turkish, a less studied language with morphologically rich nature, have demonstrated the effectiveness of neural architectures on well-formed texts and yielded state-of-the art results by formulating the task as a sequence tagging problem. In this work, we empirically investigate the use of recent neural architectures (Bidirectional long short-term memory and Transformer-based networks) proposed for Turkish NER tagging in the same setting. Our results demonstrate that transformer-based networks which can model long-range context overcome the limitations of BiLSTM networks where different input features at the character, subword, and word levels are utilized. We also propose a transformer-based network with a conditional random field (CRF) layer that leads to the state-of-the-art result (95.95\% f-measure) on a common dataset. Our study contributes to the literature that quantifies the impact of transfer learning on processing morphologically rich languages.

Domain Name System (DNS) is a hierarchical distributed naming system for computers, services, or any resource connected to the Internet. A DNS resolves queries for URLs into IP addresses for the purpose of locating computer services and devices worldwide. As of now, analytical applications with a vast amount of DNS data are a challenging problem. Clustering the features of domain traffic from a DNS data has given necessity to the need for more sophisticated analytics platforms and tools because of the sensitivity of the data characterization. In this study, a cloud based big data application, based on Apache Spark, on DNS data is proposed, as well as a periodic trend pattern based on traffic to partition numerous domain names and region into separate groups by the characteristics of their query traffic time series. Preliminary experimental results on a Turknet DNS data in daily operations are discussed with business intelligence applications.

Older adults' mental health needs significantly increased during the COVID-19 pandemic.Geriatric psychiatry is an area of extreme workforce shortage globally.A novel curriculum was developed to educate healthcare providers on COVID-19-related geriatric and geropsychiatry topics.Monthly lectures were presented from November 2020 to June 2021.Evaluations were collected after each lecture via an anonymous survey.Overall lecture quality and relevance for the participants' clinical practices were rated on a 1-3 Likert-type scale.217 participants attended lectures; 72 evaluations were collected (33% response rate).Overall lecture rating score was 2.82 ± 0.38 and relevance score was 2.77 ± 0.45.The curriculum was well received and relevant to participants.Future studies should collect more details regarding participants' clinical practices.To our knowledge, this is the first COVID-19-related geriatric psychiatry curriculum developed to educate healthcare providers and empower them to care for older adults during the pandemic.