Large-scale User Research Articles

Machine learning enabling high-throughput and remote operations at large-scale user facilities

Imaging, scattering, and spectroscopy are fundamental in understanding and discovering new functional materials. Contemporary innovations in automation and experimental techniques have led to these measurements being performed much faster and with higher resolution, thus producing vast amounts of data for analysis. These innovations are particularly pronounced at user facilities and synchrotron light sources. Machine learning (ML) methods are regularly developed to process and interpret large datasets in real-time with measurements. However, there remain conceptual barriers to entry for the facility general user community, whom often lack expertise in ML, and technical barriers for deploying ML models. Herein, we demonstrate a variety of archetypal ML models for on-the-fly analysis at multiple beamlines at the National Synchrotron Light Source II (NSLS-II). We describe these examples instructively, with a focus on integrating the models into existing experimental workflows, such that the reader can easily include their own ML techniques into experiments at NSLS-II or facilities with a common infrastructure. The framework presented here shows how with little effort, diverse ML models operate in conjunction with feedback loops via integration into the existing Bluesky Suite for experimental orchestration and data management.

Optimal Scheduling of Distribution System with Edge Computing and Data-driven Modeling of Demand Response

High penetration of renewable energies enlarge the peak-valley difference of the net load of the distribution system, which puts forward higher requirements for the operation scheduling of the distribution system. From the perspective of leveraging demand-side adjustment capabilities, an optimal scheduling method of the distribution system with edge computing and data-driven modeling of price-based demand response (PBDR) is proposed. By introducing the edge computing paradigm, a collaborative interaction framework between the control center and the edge nodes is designed for the optimization of the distribution system. At the edge nodes, a classified XG-Boost-based PBDR modeling method is proposed for large-scale differentiated users. At the control center, a two-stage optimization method integrating pre-scheduling and re-scheduling is proposed based on demand response results from all edge nodes. Through the information interaction between the control center and edge nodes, the optimized scheduling of the distribution system with large-scale users is realized. Finally, a case study is implemented on the modified IEEE 33-node system, which verifies that the proposed classified modeling method has lower errors, and it is beneficial to improve the economics of the system operation. Moreover, the simulation results show that the application of edge computing can significantly reduce the calculation time of the optimal scheduling problem with PBDR modeling of large-scale users.

Machine learning enabling high-throughput and remote operations at large-scale user facilities

Imaging, scattering, and spectroscopy are fundamental in understanding and discovering new functional materials.

IEEE Access Special Section: Privacy Preservation for Large-Scale User Data in Social Networks

Social networks have become one of the most popular platforms for people to communicate and interact with their friends and share personal information and experiences (e.g., Facebook owns over 1.23 billion monthly active users). The increasing popularity of social networks has generated extremely large-scale user data (e.g., Twitter generates 500 million tweets per day and around 200 billion tweets per year). These data can help improve people’s quality of life as well as benefit various interest groups such as advertisers, application developers, and so on. However, privacy may be compromised if learning algorithms are used to infer unpublished privacy information from published data. Hence, user data privacy preservation has become one of the most urgent research issues in social networks.

Performance enhancement of methanol reforming reactor through finned surfaces and diffused entry for on-board hydrogen generation

Hydrogen-rich combustion in engines helps in reducing pollutants significantly. But hydrogen usage on a moving vehicle is not getting large-scale user acceptance mainly due to its poor energy storage density resulting in shorter driving ranges. This storage issue led to the hunt for mediums that can efficiently produce on-board hydrogen. Methanol proves to be an efficient alcohol fuel for producing hydrogen through steam reforming reaction. The heat energy required for such endothermic reaction is obtained through exhaust engine waste energy and this process is collectively known as thermochemical recuperation. However, the conventional reactor used for this process faces a lot of problems in terms of efficiency and methanol conversion. In this study, an attempt has been made to improve the design of the reactor for on-board hydrogen generation using engine exhaust heat for addressing the challenges related to performance and hydrogen yield. For enhancing the heat transfer, a finned surface (straight & wavy) was introduced in the reactor which resulted in an increment in methanol conversion significantly. It was found that wavy fin improved the methanol conversion up to 96.8% at an exhaust inlet temperature of 673 K. Also, a diffusing inlet section was introduced to increase the residence time of reactant gases while passing through the catalyst zone. Under given inlet conditions, the methanol conversion for 6° diffuse inlet reactor goes up to 87.9% as compared to 75.4% for the conventional reactor.

Picture-based and conversational decision support to diagnose post-harvest apple diseases

This article presents the development of an expert system to support the diagnosis of post-harvest diseases of stored apples. We propose a picture-based and conversational interaction with users, where sampled images depicting symptoms of apples with known diseases are presented to users to elicit their feedback on perceived similarities in order to determine the most likely diagnosis of a diseased target apple. This article makes, besides the description of the industrial application scenario, multiple contributions circled around three rounds of user studies: (i) an usability and effectiveness assessment of the approach, where three user interface configurations are put to a test and the effectiveness of different types of user feedback mechanisms is assessed; (ii) contextual multi-armed bandit approaches for dynamic selection of displayed images with symptoms of diseased apples, that clearly outperform random and greedy sampling baseline strategies; (iii) a comparison of two different strategies for determining the context representation of a contextual multi-armed bandit approach, namely based on PCA of image features and a gamified large-scale user study. We therefore provide design insights for the development of such diagnosis applications on diseases that manifest themselves through visual symptoms in general and, hence, the findings can be also valid for domains other than post-harvest fruit diseases.

Fast-adapting and privacy-preserving federated recommender system

In the mobile Internet era, recommender systems have become an irreplaceable tool to help users discover useful items, thus alleviating the information overload problem. Recent research on deep neural network (DNN)-based recommender systems have made significant progress in improving prediction accuracy, largely attributed to the widely accessible large-scale user data. Such data is commonly collected from users’ personal devices and then centrally stored in the cloud server to facilitate model training. However, with the rising public concerns on user privacy leakage in online platforms, online users are becoming increasingly anxious over abuses of user privacy. Therefore, it is urgent and beneficial to develop a recommender system that can achieve both high prediction accuracy and strong privacy protection. To this end, we propose a DNN-based recommendation model called PrivRec running on the decentralized federated learning (FL) environment, which ensures that a user’s data is fully retained on her/his personal device while contributing to training an accurate model. On the other hand, to better embrace the data heterogeneity (e.g., users’ data vary in scale and quality significantly) in FL, we innovatively introduce a first-order meta-learning method that enables fast on-device personalization with only a few data points. Furthermore, to defend against potential malicious participants that pose serious security threat to other users, we further develop a user-level differentially private model, namely DP-PrivRec, so attackers are unable to identify any arbitrary user from the trained model. To compensate for the loss by adding noise during model updates, we introduce a two-stage training approach. Finally, we conduct extensive experiments on two large-scale datasets in a simulated FL environment, and the results validate the superiority of both PrivRec and DP-PrivRec.

A study on priming methods for graphical passwords

Recent work suggests that a type of nudge or priming technique called the presentation effect may potentially improve the security of PassPoints-style graphical passwords. These nudges attempt to prime or non-intrusively bias user password choices (i.e., point selections) by gradually revealing a background image from a particular edge to another edge at password creation time. We conduct a large-scale user study (n=710) to develop further insights into the presence of this effect and to perform the first evaluations of its security impacts. We explore the usability impacts of this effect using the subset (n=100) of participants who returned for all three sessions. Our usability analyses indicate that these priming techniques do not harm usability. Our security analyses reveal that the priming techniques can measurably enhance the security of graphical passwords; however, this effect is dependent on the combination of both the image and priming techniques used.

An approach to user knowledge acquisition in product design

As the world increasingly moves towards a knowledge-based economy, user requirements become an important factor for enterprises to drive product collaborative design evolution. To map user requirements to the product model, user requirements are generally extracted into knowledge that can be used for design decisions. However, because users are interest-driven participants and not professional design engineers, the effect of user knowledge acquisition is not ideal. There are significant challenges for rapid knowledge acquisition with dynamic user requirements. This paper presents an approach to user knowledge acquisition in the product design process, which obtains the tangible requirements of users under the premise that users are adequate for participation. In this approach, the typical information flow is divided into four stages: submission, interaction, knowledge discovery, and model evolution. In the submission stage, natural language processing technology is used to transform text form solutions into data, so that computer technology can be applied to manage large-scale user requirements. In the interaction stage, users are helped to improve their solutions by the iterative recommendation process. In the knowledge discovery stage, after less concerned partial solutions are removed and vacant items are predicted to be supplemented, the final collection of user design information is obtained. Finally, based on rough set theory, design knowledge can be extracted to support the decision of the product model. The washing machine design project is used as a case study to explain the implementation of the proposed approach.

Captivates

We present Captivates, an open-source smartglasses system designed for long-term, in-the-wild psychophysiological monitoring at scale. Captivates integrate many underutilized physiological sensors in a streamlined package, including temple and nose temperature measurement, blink detection, head motion tracking, activity classification, 3D localization, and head pose estimation. Captivates were designed with an emphasis on: (1) manufacturing and scalability, so we can easily support large scale user studies for ourselves and offer the platform as a generalized tool for ambulatory psychophysiology research; (2) robustness and battery life, so long-term studies result in trustworthy data individual's entire day in natural environments without supervision or recharge; and (3) aesthetics and comfort, so people can wear them in their normal daily contexts without self-consciousness or changes in behavior. Captivates are intended to enable large scale data collection without altering user behavior. We validate that our sensors capture useful data robustly for a small set of beta testers. We also show that our additional effort on aesthetics was imperative to meet our goals; namely, earlier versions of our prototype make people uncomfortable to interact naturally in public, and our additional design and miniaturization effort has made a significant impact in preserving natural behavior. There is tremendous promise in translating psychophysiological laboratory techniques into real-world insight. Captivates serve as an open-source bridge to this end. Paired with an accurate underlying model, Captivates will be able to quantify the long-term psychological impact of our design decisions and provide real-time feedback for technologists interested in actuating a cognitively adaptive, user-aligned future.

Leveraging End-User Data for Enhanced Design Concept Evaluation: A Multimodal Deep Regression Model

Abstract Design concept evaluation is a key process in the new product development process with a significant impact on the product's success and total cost over its life cycle. This paper is motivated by two limitations of the state-of-the-art in concept evaluation: (1) The amount and diversity of user feedback and insights utilized by existing concept evaluation methods such as quality function deployment are limited. (2) Subjective concept evaluation methods require significant manual effort which in turn may limit the number of concepts considered for evaluation. A Deep Multimodal Design Evaluation (DMDE) model is proposed in this paper to bridge these gaps by providing designers with an accurate and scalable prediction of new concepts' overall and attribute-level desirability based on large-scale user reviews on existing designs. The attribute-level sentiment intensities of users are first extracted and aggregated from online reviews. A multimodal deep regression model is then developed to predict the overall and attribute-level sentiment values based on the features extracted from orthographic product images via a fine-tuned ResNet-50 model and from product descriptions via a fine-tuned BERT model, and aggregated using a novel self-attention-based fusion model. The DMDE model adds a data-driven, user-centered loop within the concept development process to better inform the concept evaluation process. Numerical experiments on a large dataset from an online footwear store indicate a promising performance by the DMDE model with 0.001 MSE loss and over 99.1% accuracy.

ONTOLOGY-BASED APPROACH FOR FEATURE LEVEL SENTIMENT ANALYSIS

Through state-of-the-art digitalization, we can see a massive growth in user generated content on the web that provides feedback from people on a variety of topics. However, manually managing large-scale user feedback would be a difficult task and a waste of time. Therefore, the concept of sentiment analysis is emerged. Sentiment analysis is a computerized study of individuals' feelings and opinions about an entity or product. It can be executed at three different levels: document level, sentence or phrase level, and feature level. This paper proposes a novel ontology-based approach for feature level sentiment analysis. The proposed approach extracts the product features using semantic similarity and Wordnet ontology and uses the SentiWordent dictionary to classify the users’ comments as positive and negative. Furthermore, it manages negative words to obtain more precise classification results. The proposed approach is assessed by using two benchmark amazon products’ datasets in terms of accuracy; recall, precision, and f-measure. The performance reaches to 92.4% accuracy, 97.2% precision, 92.8 % recall, and 94.4% f-measure.

Redressing legal pluralism in South Africa’s water law

ABSTRACT During apartheid, South Africa formalized legal pluralism as territorial and institutional segregation that legitimized the white minority’s grab of most land, water and mineral resources. In the homelands on the remaining 13% of the country, allied tribal chiefs controlled second-class customary land and water rights. Under the democratic dispensation, the new Constitution (1996) and National Water Act (1998) aim at redressing this racial discrimination. However, these goals have not been achieved. Building on literature, field research and ongoing policy and legal debates, this paper traces causes for this failure and examines whether and how a different interpretation of statutory law can decolonize past legal pluralism. A main cause of this failure is the continued power by white large-scale water users and their consultants and lawyers. Monopolizing technical and legal knowledge, they fiercely defend apartheid era’s Existing Lawful Uses by 1998 and relatively easily obtain administrative licences for post-1998 water uptake, while claiming excessive monetary values of entitlements to water resources that the state, with tax payers, as custodian holds. In contrast, black water users’ pre-1998 Existing Lawful Uses have not been defined as yet and burdensome licences processes for new water uptake are inaccessible to the many smaller-scale black water users. The paper concludes that, in former homelands, the legal status of living customary water tenure should be elevated and protected. Inclusive facilitated processes can further clarify the “sharing in” of water resources flowing over or under the communities’ territories inside the boundaries of former homelands, also to enable gradual alignment with constitutional rights. However, in “sharing out” these water resources with powerful third parties, including foreign investors, customary water tenure should be fully protected. As a core minimum of highest priority water resource rights across the country, the current Basic Human Needs Reserve should include all water-related constitutional rights, so also water for sufficient food, and be implemented.

Time-based resource allocation for downlink in heterogeneous wireless cellular networks

Heterogeneous Wireless Cellular Networks (HWCNs) are an essential part of current and future cellular networks as a result of several benefits they offer regarding the ever-increasing user traffic requirement. Network resources are nevertheless limited, and as such, an efficient allocation of resources is vital for the development of the HWCNs. An improvement in the coverage area leads to increased mobile user satisfaction which in turn yields higher revenue for network operators. Minimising power consumption helps reduces CO $$_2$$ emissions for economically and environmentally efficient HWCNs. In this paper, by exploiting stochastic geometry, we first analyse the Coverage Probability (CP) of a typical HWCN and evaluate the impacts of propagation model/building types, cell allocation and per-tier power allocation. It is shown that a higher allocation of resources in a more lossy environment generally leads to a higher CP up to a certain threshold. Also, previous research shows that large-scale user behaviour varies over time. To this end, this paper proposes a hybrid Resource Allocation (RA) scheme, namely Time-based RA (TRA), to solve a hybrid optimisation problem of improving coverage area during periods of peak user traffic while minimising total power consumption during off-peak periods. Numerical results show that the proposed scheme achieves up to 25% higher CP during the peak period subject to limited total available power and 57% savings in total power consumption during the period of minimal user traffic given a target coverage probability, when compared to the schemes with equally allocated resources.

Beamline science – the video game: how reinforcement learning and artificial intelligence are changing operations at large-scale user facilities

Beamline science – the video game: how reinforcement learning and artificial intelligence are changing operations at large-scale user facilities

Establishing Smartphone User Behavior Model Based on Energy Consumption Data

In smartphone data analysis, both energy consumption modeling and user behavior mining have been explored extensively, but the relationship between energy consumption and user behavior has been rarely studied. Such a relationship is explored over large-scale users in this article. Based on energy consumption data, where each users’ feature vector is represented by energy breakdown on hardware components of different apps, User Behavior Models (UBM) are established to capture user behavior patterns (i.e., app preference, usage time). The challenge lies in the high diversity of user behaviors (i.e., massive apps and usage ways), which leads to high dimension and dispersion of data. To overcome the challenge, three mechanisms are designed. First, to reduce the dimension, apps are ranked with the top ones identified as typical apps to represent all. Second, the dispersion is reduced by scaling each users’ feature vector with typical apps to unit ℓ 1 norm. The scaled vector becomes Usage Pattern, while the ℓ 1 norm of vector before scaling is treated as Usage Intensity. Third, the usage pattern is analyzed with a two-layer clustering approach to further reduce data dispersion. In the upper layer, each typical app is studied across its users with respect to hardware components to identify Typical Hardware Usage Patterns (THUP). In the lower layer, users are studied with respect to these THUPs to identify Typical App Usage Patterns (TAUP). The analytical results of these two layers are consolidated into Usage Pattern Models (UPM), and UBMs are finally established by a union of UPMs and Usage Intensity Distributions (UID). By carrying out experiments on energy consumption data from 18,308 distinct users over 10 days, 33 UBMs are extracted from training data. With the test data, it is proven that these UBMs cover 94% user behaviors and achieve up to 20% improvement in accuracy of energy representation, as compared with the baseline method, PCA. Besides, potential applications and implications of these UBMs are illustrated for smartphone manufacturers, app developers, network providers, and so on.

A Research on the Matching Mechanism of Chinese Traditional Creative Products and Users: Taking Yuxian Ceramics as an Example

With the booming of creative industries in the world, handmade creative products based on Chinese traditional culture have gradually attracted the attention of academic circles. However, startups are generally faced with the problem of matching creative products with users. Existing studies have paid insufficient attention to the uniqueness of creative products and the difficulty of Product-User Fit. Based on the two-sided matching visual angle, this paper takes Yuxian Ceramics, a micro creative enterprise, as an example, and aims to reveal the diversity and hidden factors in the matching process of traditional handmade creative products and users by taking Product-User Fit as the analysis unit. The research findings are mainly as follows: (1) The disconnection between the creation and transmission of idea value is the cause of the mismatch between traditional creative products and users. (2) Build function-oriented minimal viable products (MVP) to verify the hypothesis of product function value, and find angel users. (3) Construct service-oriented MVP with product cognition and situational experience as the core, verify the hypothesis of product idea value, and attract large-scale user groups. The main contributions of this paper are as follows: (1) It identifies the dual value attributes and functions of traditional creative products. (2) It identifies the dual mechanism of matching traditional creative products with users. (3) It analyzes the interrelationship of the dual matching mechanism of matching traditional creative products with users, and enriches the entrepreneurial theory of traditional creative products. Based on the material and spiritual dual attributes of traditional creative products, two closed loops of value creation will be formed to stabilize the function value of products and expand the idea value of products. They complement each other and iterate together to promote continuous matching of creative products with users. The conclusion provides guidance for creative enterprises to explore the development of new products based on traditional cultural resources and seek two-sided matching between products and users, and has reference value for the government to develop policies and plans to foster and support entrepreneurial activities of small and micro creative enterprises.

Architecture, aesthetics, class and library attendance: An overlooked threshold for visiting public libraries

ABSTRACT Several researchers have applied Bourdieu’s concept of habitus to studies regarding the experience of space and architecture, concluding that respondents often experience surroundings very differently, depending on their social class. Therefore, it can become problematic when public institutions opt for an architectural style that could discourage members of lower/working classes to visit. We believe this to be the case for public libraries, which are housed increasingly frequent in very abstract, iconic buildings. In this article, we present different analyses into this topic performed on data that was gathered in a large scale user survey in public libraries in Flanders and a literature review that explores both the subject of iconic architecture and the relationship between class, architecture and aesthetics.

Development of Large-Scale User/Client Application for Automobile Networking Sector

Large Scale User Applications have been prevalent in the 5G era especially in sectors such as automobile, employee tracking features, e-commerce management, etc., especially with services that connect users with their pain points. One of the pain points observed in the subcontinent regarding an overlooked scenario was driving schools and the business of driving services. The current state of driving schools tends to confuse the user base with miscommunications, late service delivery, licensing formalities, and also the payment structure in the absence of a feedback loop. This project attempts to create a full-fledged driving service solution for the 38-lakh user base in the Indian Subcontinent, so as to acquire driving service providers and connect them with the target audience. This would prompt a smoother process of user onboarding as well as improve the service quality with an integrated milestone payment loop. The results observed through the launch of the application on the Play Store were positive and the young generation aged 18-15 were highly enthusiastic about using the service.

The Automated Crystallography Pipelines at the EMBL HTX Facility in Grenoble.

EMBL Grenoble operates the High Throughput Crystallization Laboratory (HTX Lab), a large-scale user facility offering high throughput crystallography services to users worldwide. The HTX lab has a strong focus in the development of new methods in macromolecular crystallography. Through the combination of a high throughput crystallization platform, the CrystalDirect technology for fully automated crystal mounting and cryocooling and the CRIMS software we have developed fully automated pipelines for macromolecular crystallography that can be remotely operated over the internet. These include a protein-to-structure pipeline for the determination of new structures, a pipeline for the rapid characterization of protein-ligand complexes in support of medicinal chemistry, and a large-scale, automated fragment screening pipeline enabling evaluation of libraries of over 1000 fragments. Here we describe how to access and use these resources.