OPTIMIZING NODE.JS APPLICATION PERFORMANCE THROUGH MAIN THREAD OFFLOADING

This paper aims to design an online, low-latency, and high-performance speech recognition system using a bidirectional long short-term memory (BLSTM) acoustic model. To achieve this, we adopt a server-client model and a context-sensitive-chunk-based approach. The speech recognition server manages a main thread and a decoder thread for each client and one worker thread. The main thread communicates with the connected client, extracts speech features, and buffers the features. The decoder thread performs speech recognition, including the proposed multichannel parallel acoustic score computation of a BLSTM acoustic model, the proposed deep neural network-based voice activity detector, and Viterbi decoding. The proposed acoustic score computation method estimates the acoustic scores of a context-sensitive-chunk BLSTM acoustic model for the batched speech features from concurrent clients, using the worker thread. The proposed deep neural network-based voice activity detector detects short pauses in the utterance to reduce response latency, while the user utters long sentences. From the experiments of Korean speech recognition, the number of concurrent clients is increased from 22 to 44 using the proposed acoustic score computation. When combined with the frame skipping method, the number is further increased up to 59 clients with a small accuracy degradation. Moreover, the average user-perceived latency is reduced from 11.71 s to 3.09–5.41 s by using the proposed deep neural network-based voice activity detector.

The user experience is the foundation for many enterprises to settle down. One of the key points to improve the user experience is to solve the user’s problems immediately and at any time. Therefore, more and more enterprises have added online customer service and other modules in the service software to respond to customer needs at any time. On the other hand, enterprises must find a balance between improving customer experience and controlling human cost. Automatic response system can achieve zero cost and 24-hour duty, which is limited to most enterprises. However, there are some problems in traditional methods, such as single guessing information, low guessing probability and poor user experience. The purpose of this paper is to provide a multi round interactive semantic analysis method based on Solr engine, which can effectively solve the above-mentioned core problems: single guessing information, low guessing probability and poor user experience. The purpose of this paper is to provide a Solr engine based multi round interactive semantic analysis method for power enterprises. In addition to effectively solving the above core problems, it can also improve the accuracy of the semantic analysis results, so that the user consultation module can get a better user experience, use the multi round interactive semantic analysis method, and output the analysis results of user emotion dimension.

Artificial Intelligence (AI) agents must interact with users across a fragmented landscape of messaging platforms, including SMS, WhatsApp, and RCS. Current agent communication strategies often rely on static preferences or simplistic fallback logic, failing to adapt to the dynamic context of interactions and the heterogeneous capabilities, costs, and reliability of available channels. This deficiency leads to suboptimal communication, increased operational costs, and poor user experience. This paper proposes the Intelligent Messaging Channel Selector (IMCS), a novel framework enabling AI agents to perform dynamic, context-aware selection of the optimal communication channel in real-time. IMCS integrates a multi-dimensional context vector-comprising message urgency, user preferences, and real-time platform performance-into a multi-criteria decision engine. Through a simulated evaluation processing <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 0 0, 0 0 0}$</tex> user interactions, IMCS demonstrated a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{4. 5 \%}$</tex> improvement in message delivery success rate and a 22 % reduction in cost per interaction compared to a standard static fallback strategy. This paper details the IMCS architecture, its decision-making logic, and the quantitative results that validate its superiority over conventional static approaches in complex, multi-channel environments.

Cloud data centers (CDCs) have become increasingly popular and widespread in recent years with the growing popularity of cloud computing and high-performance computing. Due to the multi-step computation of data streams and heterogeneous task dependencies, task failure frequently occurs, resulting in poor user experience and additional energy consumption. To reduce task execution failure as well as energy consumption, we propose a novel AI-driven energy-aware proactive fault-tolerant scheduling scheme for CDCs in this paper. First, a prediction model based on the machine learning approach is trained to classify the arriving tasks into "failure-prone tasks" and "non-failure-prone tasks" according to the predicted failure rate. Then, two efficient scheduling mechanisms are proposed to allocate two types of tasks to the most appropriate hosts in a CDC. The vector reconstruction method is developed to construct super tasks from failure-prone tasks and separately schedule these super tasks and non-failure-prone tasks to the most suitable physical host. All the tasks are scheduled in an earliest-deadline-first manner. Our evaluation results show that the proposed scheme can intelligently predict task failure and achieves better fault tolerance and reduces total energy consumption better than the existing schemes.

The increasing demand for intelligent and sustainable energy management within higher education institutions has encouraged the adoption of IoT-based solutions; however, traditional IoT dashboards typically rely on text-based device lists and non-intuitive identifiers that lack spatial context. As a result, users often struggle to understand which physical devices they are controlling, leading to confusion, poor user experience, and a higher risk of operational errors when managing smart campus facilities. This study aims to develop and validate a Digital Twin–based Smart Campus system capable of synchronizing physical electronic devices with an interactive 3D virtual environment in real time, providing a spatially accurate digital representation of the lecturer room that mirrors the real-world layout. The research employs a systematic workflow that includes problem identification, literature analysis, installation of IoT devices such as Zigbee smart switches and ESP32 IR blasters, creation of a web-based Digital Twin interface, and development of optimized 3D room models using Blender. System testing was conducted to evaluate physical-to-digital and digital-to-physical synchronization performance, and FPS benchmarking was performed to assess usability across high-end, mid-range, and entry-level devices. The results show that the Digital Twin maintains 100% synchronization accuracy with millisecond-level response times and runs smoothly on diverse hardware. By enabling users to interact with devices directly through a virtual environment that visually matches the real room, the system reduces operational mistakes, improves user experience, and enhances awareness of energy usage. The study concludes that the proposed Digital Twin approach effectively overcomes key limitations of traditional IoT dashboards and offers a scalable, practical framework for Smart Campus implementations.

The performance of data restore is one of the key indicators of user experience for backup storage systems. Compared to the traditional offline restore process, online restore reduces downtime during backup restoration, allowing users to operate on already restored files while other files are still being restored. This approach improves availability during restoration tasks but suffers from a critical limitation: inconsistencies between the access sequence and the restore sequence. In many cases, the file a user needs to access at a given moment may not yet be restored, resulting in significant delays and poor user experience. To this end, we present Histore, which builds on the user’s historical access sequence to schedule the restore sequence, in order to reduce users’ access delayed time. Histore includes three restore approaches: (i) the frequency-based approach, which restores files based on historical file access frequencies and prioritizes ensuring the availability of frequently accessed files; (ii) the graph-based approach, which preferentially restores the frequently accessed files as well as their correlated files based on historical access patterns, and (iii) the trie-based approach, which restores particular files based on both users’ real-time and historical access patterns to deduce and restore the files to be accessed in the near future. We implement a prototype of Histore and evaluate its performance from multiple perspectives. Trace-driven experiments on two datasets show that Histore significantly reduces users’ delay time by 4-700 with only 1.0%–14.5% additional performance overhead.

5G is growing globally, and the handover performance is needed to be updated and improved to adapt to new changes in telecommunication. 5G are considered small cell networks that are anticipated to have a short dwell time for users that move at high speed, like a vehicle traversing the 5G cell at a rate of 40km/h and above. It induces unnecessary handover that causes poor user experience and waste of network resources. This research tackles the problem by proposing a new handover algorithm that integrates a travel distance prediction method with AHP-TOPSIS (Analytic Hierarchy Process - Techniques for Order Preference by Similarity for an Ideal Solution) decision making. The proposed algorithm has successfully reduced the unnecessary handover in 5G networks up to 89.75% compared to the conventional TOPSIS method.

The explosion of internet news data has led to frequent information overload, which in turn has caused the dual problems of low recommendation quality and poor user experience. The research combines machine learning with image matching algorithms to provide users with interesting news content. Firstly, the news recommendation algorithm based on multi-task learning (MTL) constructs the input sequence through the users click history, maps the inputs of different tasks to the shared space and extracts text features. The image matching algorithm is adopted to record users historical preferences and capture the changing trends of interests. Experimental data show that the accuracy rate of the content-based recommendation (CB) algorithm reaches 78.27%. The collaborative filtering (CF) recommendation algorithm reached 87.97%. The recommendation accuracy of the joint recommendation algorithm all exceeds 90%. Moreover, the accuracy of the joint recommendation algorithm significantly exceeds that of the two traditional recommendation methods, CB and CF.

In the context of a quickly paced world of aging, intelligent elderly-care robots as a technology that is vital to overcome the servicing gap in relation to elderly-care services are enjoying new opportunities and challenges like never before. Based on an analysis of the current aging population and market demand, and integrating product innovation theory with commercialization promotion pathways, this paper systematically explores the product design system and market growth strategies for intelligent elderly-care robots. The study points out that current products suffer from core issues such as a disconnect between technology and demand, poor user experience, and low market penetration. It proposes a dual-track innovation system centered on "emotional interaction" and "physiological personalized services." By constructing a tiered product architecture (Tier 1–Tier 3) combined with the MVP development model, the usability, accessibility, and user acceptance of the products are significantly enhanced. Furthermore, market promotion strategies are proposed from perspectives such as policy coordination, ecological cooperation, experiential marketing, and new media communication, providing theoretical support and practical references for the widespread adoption and sustainable development of intelligent elderly-care robots.

Password-based authentication system has been praised for its user-friendly, cost-effective, and easily deployable features. It is arguably the most commonly used security mechanism for various resources, services, and applications. On the other hand, it has well-known security flaws, including vulnerability to guessing attacks. Present state-of-the-art approaches have high overheads, as well as difficulties and unreliability during training, resulting in a poor user experience and a high false positive rate. As a result, a lightweight authentication compromise detection model that can make accurate detection with a low false positive rate is required.In this paper we propose – LOG-OFF – a behavior-based authentication compromise detection model. LOG-OFF is a lightweight model that can be deployed efficiently in practice because it does not include a labeled dataset. Based on the assumption that the behavioral pattern of a specific user does not suddenly change, we study the real-world authentication traffic data. The dataset contains more than 4 million records. We use two features to model the user behaviors, i.e., consecutive failures and login time, and develop a novel approach. LOG-OFF learns from the historical user behaviors to construct user profiles and makes probabilistic predictions of future login attempts for authentication compromise detection. LOG-OFF has a low false positive rate and latency, making it suitable for real-world deployment. In addition, it can also evolve with time and make more accurate detection as more data is being collected.

With the continuous popularization of social networking sites, e.g., Twitter and Facebook, people’s requirements for the quality of social network services are rising rapidly. Traditional personalized recommendation models rely on independent algorithms, i.e., streaming ranking algorithms, content-based algorithms, and collaborative filtering analysis. However, the recommendation performance of these models for different users is not stable, resulting in poor user experience on such web services. To tackle this issue, we propose a novel recommendation model integrating multiple recommendation algorithms, which can make corresponding recommendations according to the characteristics of users. The comparative analysis with Twitter data shows that the proposed model performs better than the traditional models.KeywordsSocial networking servicesWeb user experienceUser recommendationTwitter data

One of Lehman's laws of software evolution is that the functionality of programs has to increase over time to maintain user satisfaction. In the domain of mobile apps, though, too much functionality can easily impact usability, resource consumption, and maintenance effort. Hence, does the law of continuous growth apply there? This paper shows that in mobile apps, deletion of functionality is actually common, challenging Lehman's law. We analyzed user driven requests for deletions which were found in 213,866 commits from 1,519 open source Android mobile apps from a total of 14,238 releases. We applied hybrid (open and closed) card sorting and created taxonomies for nature and causes of deletions. We found that functionality deletions are mostly motivated by unneeded functionality, poor user experience, and compatibility issues. We also performed a survey with 106 mobile app developers. We found that 78.3% of developers consider deletion of functionality to be equally or more important than the addition of new functionality. Developers confirmed that they plan for deletions. This implies the need to re-think the process of planning for the next release, overcoming the simplistic assumptions to exclusively look at adding functionality to maximize the value of upcoming releases. Our work is the first to study the phenomenon of functionality deletion and opens the door to a wider perspective on software evolution.

Currently, there is much literature discussing the smart campus framework, including terms, definitions, and practices used in campus. However, there are still many campuses facing issues with poor user experience, absence of integration platforms, inadequate data governance, etc. These obstacles have a direct impact on the overall quality and operational efficiency. Hence, a review is necessary to explore the suitable framework for smart campus. This study reviews the technology development of the smart campus and a human-centered framework is proposed to incorporate physical, cyber, and social parts to identify the essential elements of a smart campus. The proposed framework's feasibility and effectiveness are demonstrated by a real-world application. Overall, this paper contributes to the discussion on the current development of smart campuses and offers practical insights for their construction.

Keyword information retrieval is the mainstream way of information retrieval at present. Users need to scan a large amount of text information in the search results to find the information they want, but the process causes inefficiency and poor user experience. In fact, images in the web pages can provide a direct-viewing and fast retrieval way. Through the research on the features of web subject-related image and achieve its automatic identification and extraction, we can display it in the thumbnail together with the page title and summary in the results pages. It can help users filter and browse information in a more convenient way. This paper establishes a web image attribute system from both HTML attributes and external attributes. Then through corresponding automatic extracting algorithms and data analysis it succeeds to gain 16 feature rules of web subject-related image, and finishes building a web subject-related image feature model. The model proves to obtain more than 99% of the extraction rate and filtering rate while applying to the sample data, demonstrating its value in web information retrieval.

Nowadays, the quality of wireless network becomes critical for the network service providers (NSP). A poor performed network may lead to customer complaints, even loss of revenue from user churns. To ensure the quality of service, the key quality indicators (KQI) which reflect the quality of specific use cases have been collected alongside the network performance counters (NPC) for network performance analysis. To start, the NSP mines the network performance data to discover KQI anomalies which may cause poor user experience. If there is any KQI anomaly has been detected, the NSP investigates the associated NPCs to identify the possible root causes. Since the number of use cases increases dramatically and the volume of collected network performance data grows tremendously everyday, the wireless network anomaly root cause analysis becomes extremely challenging. How to efficiently discover the relationship between NPCs and KQI outliers becomes the key to identify the root causes of anomalies in the wireless network. To solve this problem, in this paper, we propose an efficient rare association rule mining algorithm called Spark-based Rare Association Rule Mining (SRAM) which leverages not only the efficiency of FP-growth algorithm but also the powerful big data processing mechanism of Spark platform. We have implemented our algorithm on the Spark platform and tested with various of data sets. The result shows our method can efficiently mine rare association rules from big volume of data.