EP-Bot: Empathetic Chatbot Using Auto-Growing Knowledge Graph

People occasionally interact with each other through conversation. In particular, we communicate through dialogue and exchange emotions and information from it. Emotions are essential characteristics of natural language. Conversational artificial intelligence is an integral part of all the technologies that allow computers to communicate like humans. For a computer to interact like a human being, it must understand the emotions inherent in the conversation and generate the appropriate responses. However, existing dialogue systems focus only on improving the quality of understanding natural language or generating natural language, excluding emotions. We propose a chatbot based on emotion, which is an essential element in conversation. EP-Bot (an Empathetic PolarisX-based chatbot) is an empathetic chatbot that can better understand a person’s utterance by utilizing PolarisX, an auto-growing knowledge graph. PolarisX extracts new relationship information and expands the knowledge graph automatically. It is helpful for computers to understand a person’s common sense. The proposed EP-Bot extracts knowledge graph embedding using PolarisX and detects emotion and dialog act from the utterance. Then it generates the next utterance using the embeddings. EP-Bot could understand and create a conversation, including the person’s common sense, emotion, and intention. We verify the novelty and accuracy of EP-Bot through the experiments.     

A Practical Approach to Constructing a Knowledge Graph for Cybersecurity

Cyberattack forms are complex and varied, and the detection and prediction of dynamic types of attack are always challenging tasks. Research on knowledge graphs is becoming increasingly mature in many fields. At present, it is very significant that certain scholars have combined the concept of the knowledge graph with cybersecurity in order to construct a cybersecurity knowledge base. This paper presents a cybersecurity knowledge base and deduction rules based on a quintuple model. Using machine learning, we extract entities and build ontology to obtain a cybersecurity knowledge base. New rules are then deduced by calculating formulas and using the path-ranking algorithm. The Stanford named entity recognizer (NER) is also used to train an extractor to extract useful information. Experimental results show that the Stanford NER provides many features and the useGazettes parameter may be used to train a recognizer in the cybersecurity domain in preparation for future work.     

中国包装产业大数据知识图谱应用系统的设计

目的针对目前包装产业存在的产业链长、数据大而散、包装领域知识不全面等问题,设计一个中国包装产业大数据知识图谱应用系统。方法从行业高度定义涵盖包装领域全生态的分类体系,结合人工智能中知识图谱的最新技术,对政府信息、工商信息、行业信息、学术论文、全球包装专利等互联网上各类数据进行自动采集汇聚,抽取出知识信息,融合成一个涵盖资讯、政策、会议、标准、论文、专利、企业、产品、高校、机构和专家等十几类信息的包装知识图谱知识库。结果系统主要功能包括数据采集、知识图谱和终端应用,实现了包装产业大数据的图谱探索、产业链图、数据报告和关联搜索。结论该系统使用方便,可从多视图、多维度获取包装产业相关数据,提升行业的数字化和信息化水平,加速中国包装行业的智能化产业升级,促进包装产业逐步向智能、绿色、集约、创新方向发展。     

Communication Mechanism and Algorithm of Composite Location Analysis of Emergency Communication based on Rail

The emergency communication system based on rail is an unconventional emergency communication mode, it is a complement equipment for that conventional communication system can’t work while tunnel mine accident occurs. Medium of transmission channel is the widely existing rail in the tunnel. In this paper we analyzed the characteristics of the rail transmission channel, verified the feasibility that information is transmitted by vibration signal in rail, we proposed the realization plan of the system. Communication protocol and processing mechanism suitable for rail transmission are designed according to the characteristics of channel bandwidth and low data transmission. Information communication with low bit rate and low bit error is realized in the communication simulation model. In the simplified model, we realized to transmit recognition speech information, and the error rate of the key text information is low to accept. The most concerned problem of personnel location in the mine disaster rescue is proposed, the composite algorithm is based on the model of signal amplitude attenuation, key node information and data frame transmission delay. Location information of hitting point can be achieved within the simplified model of the experiment. Furthermore, we discuss the characteristics of vibration signals passing through different channels.     

Alternative graph formulation for solving complex factory-scheduling problems

The adaptation of the academic job shop scheduling model to industrial practice is well known to be rich of difficulties owing to the gap between scheduling theory and practice. On the other hand, several authors observed that the disjunctive graph formulation of Roy and Sussman was more robust than the job shop model itself. In fact, the disjunctive graph can be easily adapted to deal with many practical issues. This paper moves a step closer in the direction of generalizing the existing mathematical models in order to capture relevant aspects of complex factory scheduling problems. It considers an extension of the disjunctive graph, called an alternative graph, and allows it to be modelled in a precise way to solve effectively a number of complex practical scheduling problems for which there were no successful methodologies. Several examples are presented and a complex industrial application arising in the production of steel is provided. Finally, the performance of a new fast heuristic both on real data and several instances from the literature is discussed in relation to the job shop scheduling problem.     

基于知识图谱的包装领域智能问答系统的设计

目的 针对目前包装产业存在的产业链长、数据大而散、包装领域知识信息检索不精准等问题,设计一个基于知识图谱的包装领域智能问答系统.方法 采用知识图谱、智能问答、自然语言处理、深度学习和个性化推荐等技术,对互联网上各类包装领域数据进行采集汇聚、抽取知识、融合计算,最终形成一个包装领域混合型智能问答系统.结果 系统主要功能包括知识图谱、文本相似匹配、图像识别和自动问答,实现了包装领域大数据的知识卡片、语义搜索和沉浸问答.结论 该系统使用方便,通过问答方式,一站式快速按需获取多视图、多维度包装领域数据,提升了行业的数字化、信息化和智能化水平.     

Reflection Phase Measurements for Ultrasonic NDE of Titanium Diffusion Bonds

The adoption of diffusion bonding in fracture critical titanium components has been limited by the complications that macroscopic anisotropy introduces to typical ultrasonic inspections. Previous attempts to overcome these limitations by using signal phase to extract otherwise hidden interface information showed promise but were susceptible to measurement error and proved impractical for typical aerospace component geometries. In the work presented here, significant improvements to the existing phase measurement approach are proposed alongside adaptations that permit its broader practical implementation. The principal parameters that affect the phase analysis of ultrasonic signals were investigated and their optimisation resulted in up to an order of magnitude improvement in phase measurement reliability, even at low signal-to-noise ratios. The application of these optimised parameters without a priori knowledge of the signal arrival time in an otherwise noisy waveform is illustrated, and the sensitivity of the approach to ambient temperature and annealing effects is also explored.     

Synthesis of feedback-based design concepts for sensors

We explore an approach to synthesize concepts of a class of sensors, where a quantity is sensed indirectly after nullifying its effect by using negative feedback. These sensors use negative feedback to increase the dynamic range of operation without compromising the sensitivity and resolution. The synthesis technique uses knowledge about existing phenomena to come up with an approach to synthesize concepts of sensors and also study their interactions with their surroundings, so as to generate robust designs. The approach uses a database of building blocks which are based on physical laws and effects that capture the transduction rules underlying the working principles of sensors. A simplified variant of the SAPPhIRE model of causality, which also uses physical laws and effects, has been adapted to represent the building blocks. SAPPhIRE model had been used earlier to understand analysis and synthesis of conceptual designs. We have adapted it here for automated generation of concepts. The novelty of the approach lies in the way and the ease with which it constructs a graph which is a super-set of the concept-space. The individual concepts are extracted out of the graph at a later point in time. The extraction of the concepts is done by using a modified breadth-first search algorithm which detects loops in the graph. The usage of breadth-first search algorithm for loop detection is novel, as we have demonstrated that it performs better than depth-first search algorithm for the specific problem. The technique has been implemented as a web-based application. For the sensor problems attempted, a number of existing patents were found that were based on the concepts that were generated by the synthesis algorithm, thus emphasizing the usefulness of the designs produced. The tool generated 35 concepts for accelerometers, out of which 2 concepts were found in patents. The synthesis approach also proposed new, feasible sensor concepts, thereby indicating its potential as a stimulator for enhancing creativity of designers. Automated generation of feedback-based sensor designs is a novel outcome of this approach.     

基于因特网的知识管理工具的设计

基于因特网的知识管理工具能够支持企业有效、快速地识别、获取、开发、分解、储存和传递知识 ,从而使每个员工在最大限度地贡献出其积累的知识的同时 ,也能享用他人的知识 ,实现知识共享 .本文主要介绍了一个基于因特网知识管理工具的设计思想、框架及所用的知识管理模型等 .知识管理模型包括知识结构图、知识利用图、知识分布图以及知识传递图等     

Entity and relation extraction with rule-guided dictionary as domain knowledge

Entity and relation extraction is an indispensable part of domain knowledge graph construction, which can serve relevant knowledge needs in a specific domain, such as providing support for product research, sales, risk control, and domain hotspot analysis. The existing entity and relation extraction methods that depend on pretrained models have shown promising performance on open datasets. However, the performance of these methods degrades when they face domain-specific datasets. Entity extraction models treat characters as basic semantic units while ignoring known character dependency in specific domains. Relation extraction is based on the hypothesis that the relations hidden in sentences are unified, thereby neglecting that relations may be diverse in different entity tuples. To address the problems above, this paper first introduced prior knowledge composed of domain dictionaries to enhance characters’ dependence. Second, domain rules were built to eliminate noise in entity relations and promote potential entity relation extraction. Finally, experiments were designed to verify the effectiveness of our proposed methods. Experimental results on two domains, including laser industry and unmanned ship, showed the superiority of our methods. The F1 value on laser industry entity, unmanned ship entity, laser industry relation, and unmanned ship relation datasets is improved by +1%, +6%, +2%, and +1%, respectively. In addition, the extraction accuracy of entity relation triplet reaches 83% and 76% on laser industry entity pair and unmanned ship entity pair datasets, respectively.     

Factor graph methods for three-dimensional shape reconstruction as applied to LIDAR imaging

Two methods based on factor graphs for reconstructing the three-dimensional (3D) shape of an object from a series of two-dimensional images are presented. First, a factor graph model is developed for image segmentation to obtain silhouettes from raw images; the shape-from-silhouette technique is then applied to yield the 3D reconstruction of the object. The second method presented is a direct 3D reconstruction of the object using a factor graph model for the voxels of the reconstruction. While both methods should be applicable to a variety of input data types, they will be developed and demonstrated for a particular application involving the LIDAR imaging of a submerged target. Results from simulations and from real LIDAR data are shown that detail the performance of the methods.     

Big Data Analytics Using Graph Signal Processing

The networks are fundamental to our modern world and they appear throughout science and society. Access to a massive amount of data presents a unique opportunity to the researcher’s community. As networks grow in size the complexity increases and our ability to analyze them using the current state of the art is at severe risk of failing to keep pace. Therefore, this paper initiates a discussion on graph signal processing for large-scale data analysis. We first provide a comprehensive overview of core ideas in Graph signal processing (GSP) and their connection to conventional digital signal processing (DSP). We then summarize recent developments in developing basic GSP tools, including methods for graph filtering or graph learning, graph signal, graph Fourier transform (GFT), spectrum, graph frequency, etc. Graph filtering is a basic task that allows for isolating the contribution of individual frequencies and therefore enables the removal of noise. We then consider a graph filter as a model that helps to extend the application of GSP methods to large datasets. To show the suitability and the effeteness, we first created a noisy graph signal and then applied it to the filter. After several rounds of simulation results. We see that the filtered signal appears to be smoother and is closer to the original noise-free distance-based signal. By using this example application, we thoroughly demonstrated that graph filtration is efficient for big data analytics.     

Engagement Detection Based on Analyzing Micro Body Gestures Using 3D CNN

This paper proposes a novel, efficient and affordable approach to detect the students’ engagement levels in an e-learning environment by using webcams. Our method analyzes spatiotemporal features of e-learners’ micro body gestures, which will be mapped to emotions and appropriate engagement states. The proposed engagement detection model uses a three-dimensional convolutional neural network to analyze both temporal and spatial information across video frames. We follow a transfer learning approach by using the C3D model that was trained on the Sports-1M dataset. The adopted C3D model was used based on two different approaches; as a feature extractor with linear classifiers and a classifier after applying fine-tuning to the pre-trained model. Our model was tested and its performance was evaluated and compared to the existing models. It proved its effectiveness and superiority over the other existing methods with an accuracy of 94%. The results of this work will contribute to the development of smart and interactive e-learning systems with adaptive responses based on users’ engagement levels.     

Above-Extreme Extraction Mode in the Purex Process: 1. Calculation of the Pu and U Distribution in the Head Extractor

Accumulation of Pu(IV) in head extractors of the mixer-settler type as applied to two different flowsheets of NPP spent fuel reprocessing was simulated under dynamic and steady-state conditions using a mathematical model of extraction. Comparison of the calculations showed that the maximal accumulation of Pu in the head extractor is independent of the calculation procedure and is determined by the features of the extractor flowsheet. However, dangerous accumulation of Pu according to calculations for steady-state conditions (extreme mode of extraction) occurs at smaller deviations of the flow rates from the nominal values, as compared to calculations for the dynamic conditions (above-extreme extraction mode). Accumulation of Pu to high concentrations under steady-state conditions requires a longer time to attain the equilibrium, whereas under dynamic conditions Pu is accumulated considerably faster.     

Relative positioning of a load extractor for a storage carousel

The use of carousels in manufacturing for the storage of work-in-process items is very common. A major reason for the popular use of carousels is to store work-in-process items closer to the workstations and therefore minimize handling and the time taken to satisfy an item retrieval request from storage. Although carousel load retrieval times are considerably less than those of normal warehouses, the average response time can be further improved through proper prepositioning of the load extractor machine in anticipation of storage/retrieval requests. Two models are presented in this paper to strategically preposition the extractor machine when idle to minimize the system response time. In the first model the emphasis is on preposition of the extractor machine to minimize the maximum system response time when the extractor machine becomes idle. In contrast, the emphasis on the second model is to preposition the machine to minimize the expected system response time when the machine becomes idle. Two modes of carousel operation, (a) unidirectional rotation capability and (b) bidirectional rotation capability, are considered. Solution procedures based on mathematical models are developed to obtain optimal solutions. The developed procedures can be embedded in a carousel controller and used for on-line control and prepositioning of the extractor machine to improve the carousel system's operational efficiency. Computer simulation is performed under various traffic rates and storage methods to analyze the system performance under these two prepositioning strategies and one other traditional carousel prepositioning strategy commonly in use today. An on-line control architecture to implement the proposed prepositioning strategies is also presented.     

Modeling Multi-Targets Sentiment Classification via Graph Convolutional Networks and Auxiliary Relation

Existing solutions do not work well when multi-targets coexist in a sentence. The reason is that the existing solution is usually to separate multiple targets and process them separately. If the original sentence has N target, the original sentence will be repeated for N times, and only one target will be processed each time. To some extent, this approach degenerates the fine-grained sentiment classification task into the sentencelevel sentiment classification task, and the research method of processing the target separately ignores the internal relation and interaction between the targets. Based on the above considerations, we proposes to use Graph Convolutional Network (GCN) to model and process multi-targets appearing in sentences at the same time based on the positional relationship, and then to construct a graph of the sentiment relationship between targets based on the difference of the sentiment polarity between target words. In addition to the standard target-dependent sentiment classification task, an auxiliary node relation classification task is constructed. Experiments demonstrate that our model achieves new comparable performance on the benchmark datasets: SemEval-2014 Task 4, i.e., reviews for restaurants and laptops. Furthermore, the method of dividing the target words into isolated individuals has disadvantages, and the multi-task learning model is beneficial to enhance the feature extraction ability and expression ability of the model.     

Modelling lung cancer due to radon and smoking in WISMUT miners: preliminary results

A mechanistic two-stage carcinogenesis model has been applied to model lung-cancer mortality in the largest uranium-miner cohort available. Models with and without smoking action both fit the data well. As smoking information is largely missing from the cohort data, a method has been devised to project this information from a case-control study onto the cohort. Model calculations using 256 projections show that the method works well. Preliminary results show that if an explicit smoking action is absent in the model, this is compensated by the values of the baseline parameters. This indicates that in earlier studies performed without smoking information, the results obtained for the radiation parameters are still valid. More importantly, the inclusion of smoking-related parameters shows that these mainly influence the later stages of lung-cancer development.     

Analysing source code: looking for useful verb?direct object pairs in all the right places

The large time and effort devoted to software maintenance can be reduced by providing software engineers with software tools that automate tedious, error-prone tasks. However, despite the prevalence of tools such as IDEs, which automatically provide program information and automated support to the developer, there is considerable room for improvement in the existing software tools. The authors' previous work has demonstrated that using natural language information embedded in a program can significantly improve the effectiveness of various software maintenance tools. In particular, precise verb information from source code analysis is useful in improving tools for comprehension, maintenance and evolution of object-oriented code, by aiding in the discovery of scattered, action-oriented concerns. However, the precision of the extraction analysis can greatly affect the utility of the natural language information. The approach to automatically extracting precise natural language clues from source code in the form of verb- direct object (DO) pairs is described. The extraction process, the set of extraction rules and an empirical evaluation of the effectiveness of the automatic verb-DO pair extractor for Java source code are described.     

基于贝叶斯网络的系统测试性建模与分析

针对现有测试性建模方法对系统不确定信息描述及分析上的不足,提出了基于贝叶斯网络的测试性建模与分析方法。首先结合贝叶斯网络的基本理论,阐述了系统测试性模型的构成要素及建立流程,用相关性矩阵表示系统故障-测试之间的关联,用条件概率矩阵描述两者间的不确定信息;然后给出了测试性指标的具体分析过程和算法;最后通过实例验证了方法的有效性。