结合句法依存信息的方面级情感分类

引入句法依存信息到原方面术语,提出一种新的方面术语表示方法,利用Glove词向量表示单词以及单词与单词之间的依存关系,构造出包含句法依存信息的依存关系邻接矩阵和依存关系表示矩阵,利用图卷积神经网络和多头注意力机制将句法依存信息融入到方面术语中,使得方面术语表达与上下文结构高度相关。将改进后的方面词术语表示替换到现有模型后,模型泛化能力得到有效提升。对比试验和分析结果表明:该方法具有有效性和泛化性。     

Morphological study of thin films: Simulation and experimental insights using horizontal visibility graph

We studied the morphological nature of various thin films such as silicon carbide (SiC), diamond (C), germanium (Ge), and gallium nitride (GaN) on silicon substrate Si(100) using the pulsed laser deposition (PLD) method and Monte Carlo simulation. We, for the first time, systematically employed the visibility algorithm graph to meticulously study the morphological features of various PLD grown thin films. These thin-film morphologies are investigated using random distribution, Gaussian distribution, patterned heights, etc. The nature of the interfacial height of individual surfaces is examined by a horizontal visibility graph (HVG). It demonstrates that the continuous interfacial height of the silicon carbide, diamond, germanium, and gallium nitride films are attributed to random distribution and Gaussian distribution in thin films. However, discrete peaks are obtained in the brush and step-like morphology of germanium thin films. Further, we have experimentally verified the morphological nature of simulated silicon carbide, diamond, germanium, and gallium nitride thin films were grown on Si(100) substrate by pulsed laser deposition (PLD) at elevated temperature. Various characterization techniques have been used to study the morphological, and electrical properties which confirmed the different nature of the deposited films on the Silicon substrate. Decent hysteresis behavior has been confirmed by current-voltage (IV) measurement in all the four deposited films. The highest current has been measured for GaN at ～60 nA and the lowest current in SiC at ～30 nA level which is quite low comparing with the expected signal level (μA). The HVG technique is suitable to understand surface features of thin films which are substantially advantageous for the energy devices, detectors, optoelectronic devices operating at high temperatures.     

基于人脸卡口数据的行人共现关系图谱构建

随着人脸识别技术的不断进步以及人脸卡口的大范围且密集的部署, 本文针对团伙犯罪案件侦察这一应用场景, 对人脸卡口数据进行深入挖掘, 探究其中行人间的共现关系, 获取所关注的嫌疑人的现实社交网络, 锁定团伙其余人员. 经过实验比对和论证, 本文使用Chinese Whispers聚类算法对行人节点进行识别, 通过Faiss加速邻接边的构建, 加速图的初始化步骤, 解决其聚类效率低下的问题. 在此基础上, 使用共现频次和Apriori算法中的置信度挖掘行人间的共现关系, 构建行人共现关系图谱.     

LinkingPark: An automatic semantic table interpretation system

In this paper, we present LinkingPark, an automatic semantic annotation system for tabular data to knowledge graph matching. LinkingPark is designed as a modular framework which can handle Cell-Entity Annotation (CEA), Column-Type Annotation (CTA), and Columns-Property Annotation (CPA) altogether. It is built upon our previous SemTab 2020 system, which won the 2nd prize among 28 different teams after four rounds of evaluations. Moreover, the system is unsupervised, stand-alone, and flexible for multilingual support. Its backend offers an efficient RESTful API for programmatic access, as well as an Excel Add-in for ease of use. Users can interact with LinkingPark in near real-time, further demonstrating its efficiency.     

电商平台产品共同购买网络的影响因素研究

共同购买网络的推荐系统应用越加广泛，仅基于网络内生结构变量研究其共同购买链接的经济意义已有局限，故加入网络口碑这一外生变量，进行更全面的分析。采用了社会网络方法中指数随机图模型进行建模，主要围绕产品销售量、产品入度、差评率和评论数4个方面因素，探究其对共同购买网络中共同购买链接形成的影响。结果显示，销售量、产品入度和评论数对共同购买链接形成的影响呈正比关系，而差评率则会显著地降低产品共同购买的几率。该指数随机图构建出的共同购买网络为在线电商平台管理网络口碑和推荐系统优化设计提供有益参考。     

一种基于GN算法的动态图划分方法

随着图规模的急剧增长，对动态图进行实时处理的需求日益增加。大多现有的算法针对静态图划分是有效的，直接用其处理动态图会带来较大的通信开销。针对该问题，提出一种基于GN算法的动态图划分方法。首先收集一段时间内加入动态图中的顶点；然后，利用GN算法对这些新加入的顶点进行预划分，产生若干个内部联系紧密的社区；最后，将预划分产生的社区结果插入到已经划分好的当前图中。实验从交叉边数和负载均衡度两方面将该方法与传统流式划分方法进行比较，结果表明， 在公开数据集上，该方法的交叉边数降低了13%，负载均衡度减少了42.3%。由此可见，该方法的划分质量明显优于传统的流式划分方法。     

一种使用多跳事实的端到端知识库实体描述生成方法

自动化实体描述生成有助于进一步提升知识图谱的应用价值,而流畅度高是实体描述文本的重要质量指标之一。该文提出使用知识库上多跳的事实来进行实体描述生成,从而贴近人工编撰的实体描述的行文风格,提升实体描述的流畅度。该文使用编码器—解码器框架,提出了一个端到端的神经网络模型,可以编码多跳的事实,并在解码器中使用关注机制对多跳事实进行表示。该文的实验结果表明,与基线模型相比,引入多跳事实后模型的BLEU-2和ROUGE-L等自动化指标分别提升约8.9个百分点和7.3个百分点。     

OWL‐based acquisition and editing of computer‐interpretable guidelines with the CompGuide editor

Computer‐Interpretable Guidelines (CIGs) are the dominant medium for the delivery of clinical decision support, given the evidence‐based nature of their source material. Therefore, these machine‐readable versions have the ability to improve practitioner performance and conformance to standards, with availability at the point and time of care. The formalisation of Clinical Practice Guideline knowledge in a machine‐readable format is a crucial task to make it suitable for the integration in Clinical Decision Support Systems. However, the current tools for this purpose reveal shortcomings with respect to their ease of use and the support offered during CIG acquisition and editing. In this work, we characterise the current landscape of CIG acquisition tools based on the properties of guideline visualisation, organisation, simplicity, automation, manipulation of knowledge elements, and guideline storage and dissemination. Additionally, we describe the CompGuide Editor, a tool for the acquisition of CIGs in the CompGuide model for Clinical Practice Guidelines that also allows the editing of previously encoded guidelines. The Editor guides the users throughout the process of guideline encoding and does not require proficiency in any programming language. The features of the CIG encoding process are revealed through a comparison with already established tools for CIG acquisition.     

Topological Characterization of Book Graph and Stacked Book Graph

Degree based topological indices are being widely used in computer-aided modeling, structural activity relations, and drug designing to predict the underlying topological properties of networks and graphs. In this work, we compute the certain important degree based topological indices like Randic index, sum connectivity index, ABC index, ABC₄ index, GA index and GA₅ index of Book graph B_n and Stacked book graph B_m,n. The results are analyzed by using edge partition, and the general formulas are derived for the above-mentioned families of graphs.