Search

Although the development of AI in games is remarkable, intelligent machines still lag behind humans in games that require the ability of language understanding. In this paper, we focus on the crossword puzzle resolution task. Solving crossword puzzles is a challenging task since it requires the ability to answer natural language questions with knowledge and the ability to execute a search over possible answers to find an optimal set of solutions for the grid. Previous solutions are devoted to exploiting heuristic strategies in search to find solutions while having limited ability to explore the search space. We build a comprehensive system for crossword puzzle resolution based on Monte Carlo Tree Search (MCTS). As far as we know, we are the first to model the crossword puzzle resolution problem as a Markov Decision Process and apply the MCTS to solve it. We construct a dataset for crossword puzzle resolution based on daily puzzles from The New York Times with detailed specifications of both the puzzle and clue database selection. Our method achieves state-of-the-art performance on the dataset. The code of the system and experiments in this paper is publicly available: https://www.github.com/lhlclhl/CP. [ABSTRACT FROM AUTHOR]

In the complex field of radiotherapy (RT), ensuring patient safety is paramount. Recognizing the critical need for disseminating knowledge on potential failure modes (FMs) in RT, our study introduces i-SART (intelligent Safety Assistant in RT), a web application empowered by AI, aiming to promote proactive risk management and patient safety awareness. The objective of this work is to describe i-SART's development and deployment strategy and discuss its potential implications. The i-SART design is based on the Failure Modes and Effects Analysis methodology (FMEA) and the AAPM TG100, whilst integrating AI techniques to engage RT professionals in an interactive learning process. I-SART is built upon a database populated with potential FMs in the RT process and associated techniques, including MR-guided Adaptive RT, extracted from 10 papers and safety reports. The application, developed using Python 3.9 and the Django REST framework, features two user roles: administrator and user. Secure information transfer was ensured through JSON Web Token authentication. MySQL relational database was chosen for persistent user and FM information storage. The user interface of i-SART was created using Vue.js, an interpreted computer programming language-based framework. An AI-driven chatbot facilitates discussions about FMs and safety measures. Experiments with AI techniques such as Markov Chains and Generative Adversarial Networks were conducted to generate synthetic FMs. This is a collaborative effort between two European Universities. The application collects data and organizes a database of potential FMs and their attributes (causes, effects, severity and mitigation strategies) in RT procedures. Initially, 728 FMs were collected which were then reduced to 419 after the removal of duplicates and ambiguities. The application provides users with functionalities for searching, filtering, and sorting FMs, as well as contributing new FMs for continuous database growth. The AI-driven chatbot is anticipated to enhance user engagement and knowledge dissemination. The experiments with AI techniques for synthetic FMs resulted in limited outcomes due to the small training dataset. I-SART's effectiveness will soon be evaluated by users and stakeholders. I-SART is a promising conversational tool for enhancing patient safety in RT through proactive risk management. Future work involves expanding the database and exploring integration with other risk management methodologies. We anticipate i-SART to drive further research in integrating AI techniques in RT risk management and providing a substantial dataset for machine learning applications, FMs predictions and synthetic FMs generation. [ABSTRACT FROM AUTHOR]

Over the past year, ChatGPT's initial applications across diverse fields have progressed amid debates on its ethical implications and inherent biases. This paper investigates the comprehension level of ChatGPT 4.0 regarding cementitious composites and initiates preliminary research into its application in cementitious composites，including ordinary Portland cement (OPC) concrete, Engineered Cementitious Composite (ECC), and Ultra-High-Performance Concrete (UHPC). Initially, the study designed forty-five progressively challenging questions to examine the readability and accuracy of ChatGPT's responses related to cementitious composites. Building on this, the research explored the feasibility of utilizing ChatGPT for the identification and analysis of ECC cracks, the design of mix proportions for cementitious composites, and the real-time analysis of carbon emissions. The results demonstrate ChatGPT 4.0's proficiency in acquiring foundational knowledge of cementitious composites, a capability verified through interactive human-computer modalities for analysing crack quantity and width, formulating preliminary material compositions, and providing instantaneous feedback on carbon emissions. However, the study also uncovers ChatGPT 4.0's limitations in understanding the complex design mathematics and mechanics underlying cementitious composites, as well as its inadequate ability to synthesise and integrate interdisciplinary knowledge, occasionally leading to arbitrary and forced responses. These findings underscore the need for further enhancements in algorithmic sophistication and database comprehensiveness. The paper concludes by projecting the future applicability of ChatGPT in the cementitious composites sector, highlighting its initial contributions while delineating avenues for subsequent research and development. • The understanding of ChatGPT on cementitious composites was investigated. • The preliminary application study of ChatGPT 4.0 was conducted. • The future application prospects of ChatGPT were anticipated. [ABSTRACT FROM AUTHOR]

The decision-making on green retrofitting of existing buildings relies on both explicit and implicit knowledge, and its efficiency and reliability need improvement. Intelligent approaches that can sufficiently utilize the text information of existing projects are required to provide more suitable strategies for green retrofitting. This paper describes a decision-making approach combining Case-Based Reasoning (CBR) and Random Forest (RF), which can identify similar cases from the database containing 109 green retrofit projects and revise outdated measures. A practical project case study shows that the revised retrofit measures can reduce Energy Use Intensity (EUI) by 37%. The proposed approach optimizes and standardizes CBR processes and provides guidance for coping with semi-structured green retrofit decision-making problems, thereby promoting sustainable development and intelligent management in the construction field. The system prototype will be developed and promoted after the case database is expanded. • A CBR-RF approach is designed to select more energy-efficient strategies for building green retrofit. • RF is proven to optimize the case revision stage of CBR to avoid outdated retrofit measures. • The information on existing green retrofit projects is standardized and cross-platform used. • This approach provides intelligent guidance for retrofit decision-making, improving efficiency and reliability. [ABSTRACT FROM AUTHOR]

Seamless positioning ability has become an essential requirement in large-scale smart city scenes with the development of Artificial Intelligence of Things technology. The performance of seamless positioning is limited by the inaccurate crowdsourced navigation database, cumulative error of built-in sensors, and changeable measurement errors of different location sources. In order to solve these problems, this paper presents the CrowdLOC-S framework, which provides a concrete and accurate indoor/outdoor localization performance using the combination of crowdsourced Wi-Fi fingerprinting, Global Navigation Satellite System (GNSS), and low-cost sensors. A data and model dual-driven based trajectory estimator is developed for improving the long-term positioning performance of built-in sensors, and a hybrid one-dimensional convolutional neural network (1D-CNN), Bi-directional Long Short-Term Memory (Bi-LSTM), and Multilayer Perceptron (MLP) enhanced quality indicator is proposed for quality evaluation of crowdsourced trajectories and further Wi-Fi fingerprinting database construction. Besides, the transfer learning approach is applied in the quality indicator for autonomously predicting the location errors towards different indoor and outdoor location sources and realizing seamless scenes switching. Finally, a unified extended Kalman filter is developed to realize multi-source integration-based seamless localization using the positioning information provided by indoor and outdoor location sources and corresponding quality indicator results. Comprehensive experiments demonstrate that the presented CrowdLOC-S system is proven to realize precise and efficient indoor and outdoor positioning performance in complex and large-scale urban environments. [ABSTRACT FROM AUTHOR]

• Artificial Intelligence to predict two-phase frictional pressure gradient. • Collection of a pressure drop database for a wide range of operating conditions. • Effect of the output and input type and numerosity on prediction accuracy. • Good accuracy with non-dimensional output and mixed input values. • Conventional methods always preserve physical trend with respect to neural networks. This paper presents a discussion on several possibilities to predict the frictional pressure gradient during two-phase flow, with both the application of artificial intelligence and the implementation of conventional correlations and predictive methods. To this purpose, a huge database of approximately 8000 data points has been collected from 49 sources available in scientific literature, including 23 working fluids and the following ranges of parameters: mass fluxes from 32.7 to 2000 kg/m2s, saturation temperatures from -190°C to +120°C (reduced pressures from 0.021 to 0.780), tube diameters from 0.5 to 14.0 mm. This consolidated database has been used to train several artificial neural networks (ANNs), by using only two hidden layers (shallow neural networks) and evaluating the effect of: training and testing datasets choice (either test data included or outside the training domain), the number of neurons for each hidden layer (from 1 to 50), the type of output (either dimensional or non-dimensional), the type and number (from 1 to 22) of input parameters. The best results (MAPE of 16.8% and 88% of data within ±30%) have been obtained by using the liquid-only two-phase multiplier Φ L O 2 as non-dimensional output and 12 mixed input parameters. Compared to the statistics of well-established literature correlations for frictional pressure drop (best MAPE of 22% and 73% of data points predicted within a ±30% error range, provided by Mauro et al. mechanistic method), the ANN demonstrates therefore a higher general accuracy. However, the use of Artificial Neural Networks does not guarantee a physical trend, which is instead preserved with conventional prediction methods. [ABSTRACT FROM AUTHOR]

Determining frequency of buildings has a crucial impact on proper analysis and design of structures. Conventional methods (i.e., in situ and numerical analysis) are generally costly and time-consuming. This study aims at developing Machine Learning (ML)-based methods for obtaining frequency of structures with the focus on masonry towers (i.e., bell towers). To this end, a database including the results of either in situ or numerical analysis on over 90 masonry towers available in literature is collected. Additionally, frequencies of 18 towers located in Venice, Italy are measured by site survey, and added to the database. Parameters with the highest influence on tower's frequency, namely height, plan dimensions and modulus of elasticity are defined as input variables for predicting natural frequency of a tower by ML-based techniques including Decision Tree (DT), Random Forest (RF), XGBoost and K-Nearest Neighbors (KNN). The models' performance is analyzed by comparing the correlation between the predicted and real values. Moreover, the models' accuracy is assessed through common performance metrics and Taylor diagram, and the most accurate model is introduced accordingly. Results highlighted the high capability of ML-based approaches for predicting towers' frequency, and the XGBoost model exhibited the highest accuracy. In the second part of the paper, the values predicted by the most accurate model are compared to those calculated by the equations proposed by design Codes (i.e., NTC008, NCSE, ASCE) and literature. Lastly, for deeper investigating the performance of masonry towers a sensitivity analysis is performed by the proposed XGBoost model, and an equation is suggested for calculating their natural frequency based on their height and plan width. • A database of the natural frequency of towers is collected from the literature as well as on-site surveys. • ML-based models are proposed for predicting the natural frequency of towers. • Accuracy of the ML-based models is compared to that of equations available in design codes. • An equation is proposed for calculating natural frequency of towers based on their height and plan width. [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) is the latest designing and interacting technology used to support organizations in the competitive electronic commerce (E-commerce) context. The studies on AI in E-commerce remain fragmentary since few papers review the evolutionary process of knowledge infrastructure. To bridge this gap, we constructed a conceptual framework from the resource orchestration perspective, and retrieved 2252 documents from the Web of Science (WoS) database dating from 1998 to 2022 to synthesize the extant research on AI in E-commerce. Specifically, we integrated main path analysis with science mapping analysis covering the strategic diagram and evolution map. Through qualitative analysis of critical nodes in main paths, this article found that knowledge disseminated around recommender systems (RSs). This article also examined the performance traits and inter-relationships of significant themes over three consecutive time zones, as well as identified four main AI-enabled outcomes (thematic areas): Optimization and decision support , Trust and personalized recommendation , Sentiment analysis, and AI theories with allied technologies. Potential research prospects and implications are proposed based on these findings. Overall, this article provides state-of-the-art information on how AI can facilitate the implementation of E-commerce operations and could serve as a stepping stone to newcomers in the scientific community. [ABSTRACT FROM AUTHOR]

The purpose of this paper is to systematically review research related to the value of new technologies in the context of performance measurement and management. The review synthesizes findings from previous studies to provide a comprehensive understanding of the current state-of-the-art of various novel technologies and how they assist performance measurement and management across various industries. Scopus was selected as the main electronic database for this study. As a principal result, we found that research related to the value of novel technologies in the context of performance measurement and management is formed around six research streams: automation techniques, predictive analytics, real-time data collection, visualization, collaboration tools, and blockchain technology. As a conclusion, the research presents that novel technologies can be used to track and monitor performance, automate processes, and streamline operations. The findings of this review will bring more coherent value to the current state of the literature and contribute to a diverse body of knowledge. [ABSTRACT FROM AUTHOR]