改进的多层感知机在客户流失预测中的应用

针对在传统的客户流失预测数据预处理中，使用one-hot编码处理离散属性导致数据维度增加及数据过于稀疏的问题，提出了两种基于多层感知机的改进后的客户流失预测模型。其主要思想是分别使用堆叠自编码器和实体嵌入两种方法对多层感知机进行改进，通过将离散属性的高维编码数据向低维空间映射，有效地减少了one-hot编码产生的稀疏数据，增加了离散属性值之间的关联度。在对两份公开的数据集进行交叉验证后的实验结果表明，改进后的模型既有效地提高了预测的准确度，又维持了传统多层感知机模型在并行化计算方面的优势。     

命名实体识别技术综述

命名实体识别是自然语言处理中的热点研究方向之一,目的是识别文本中的命名实体并将其归纳到相应的实体类型中。首先阐述了命名实体识别任务的定义、目标和意义,分析提出了命名实体识别的主要难点在于领域命名实体识别局限性、命名实体表述多样性和歧义性、命名实体的复杂性和开放性;然后介绍了命名实体识别研究的发展进程,从最初的规则和字典方法到传统的统计学习方法再到现在的深度学习方法,不断地将新技术应用到命名实体识别研究中以提高性能;接着系统梳理了当下命名实体识别任务中的若干热门研究点,分别是匮乏资源下的命名实体识别、细粒度命名实体识别、嵌套命名实体识别以及命名实体链接;最后针对评判命名实体识别模型的好坏,总结了常用的若干数据集和实验测评指标,并给出了未来的研究建议。     

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.     

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。     

利用二次归并的Deep Web实体匹配方法

针对权重边剪枝（WEP）方法在准确率和匹配效率等方面的不足，通过引入自匹配和归并概念，提出一种基于二次归并的Deep Web实体匹配方法。首先，提取各对象的属性值，并按属性值重组对象，使具有相同属性值的对象聚集在一起，实现块的有效划分；其次，计算块内各对象间的匹配度，并据此进行剪枝、自匹配检测、归并，输出初步类簇；最后，以初步类簇为基础，利用簇内对象间传递的消息以及对象属性相似值，进一步挖掘匹配关系，触发新一轮的类簇归并与更新。实验结果表明，与WEP方法相比，所提方法通过自匹配检测，自动区分匹配关系并采取合适的匹配策略，使归并过程逐渐精化，提高了匹配准确率；通过分块、剪枝，有效缩减了匹配空间，提高了系统运行效率。     

健康领域Web信息抽取

针对Web信息抽取(WIE)技术在健康领域应用的问题,提出了一种基于WebHarvest的健康领域Web信息抽取方法。通过对不同健康网站的结构分析设计健康实体的抽取规则,实现了基于WebHarvest的自动抽取健康实体及其属性的算法;再把抽取的实体及其属性进行一致性检查后存入关系数据库中,然后对关系数据库中隐含健康实体的属性值利用Ansj自然语言处理方法进行实体识别, 进而抽取健康实体之间的联系。该技术在健康实体抽取实验中,平均F值达到99.9%,在实体联系抽取实验中,平均F值达到80.51%。实验结果表明提出的Web信息抽取技术在健康领域抽取的健康信息具有较高的质量和可信性。     

Enriching Context Information for Entity Linking with Web Data

Journal of Computer Science and Technology - Entity linking (EL) is the task of determining the identity of textual entity mentions given a predefined knowledge base (KB). Plenty of existing...     

DEM: Deep Entity Matching Across Heterogeneous Information Networks

Heterogeneous information networks, which consist of multi-typed vertices representing objects and multi-typed edges representing relations between objects, are ubiquitous in the real world. In this paper, we study the problem of entity matching for heterogeneous information networks based on distributed network embedding and multi-layer perceptron with a highway network, and we propose a new method named DEM short for Deep Entity Matching. In contrast to the traditional entity matching methods, DEM utilizes the multi-layer perceptron with a highway network to explore the hidden relations to improve the performance of matching. Importantly, we incorporate DEM with the network embedding methodology, enabling highly efficient computing in a vectorized manner. DEM’s generic modeling of both the network structure and the entity attributes enables it to model various heterogeneous information networks flexibly. To illustrate its functionality, we apply the DEM algorithm to two real-world entity matching applications: user linkage under the social network analysis scenario that predicts the same or matched users in different social platforms and record linkage that predicts the same or matched records in different citation networks. Extensive experiments on real-world datasets demonstrate DEM’s effectiveness and rationality.

     

A novel approach for efficient stance detection in online social networks with metaheuristic optimization

In the 19th and 20th centuries, social networks have been an important topic in a wide range of fields from sociology to education. However, with the advances in computer technology in the 21st century, significant changes have been observed in social networks, and conventional networks have evolved into online social networks. The size of these networks, along with the large amount of data they generate, has introduced new social networking problems and solutions. Social network analysis methods are used to understand social network data. Today, several methods are implemented to solve various social network analysis problems, albeit with limited success in certain problems. Thus, the researchers develop new methods or recommend solutions to improve the performance of the existing methods. In the present paper, a novel optimization method that aimed to classify social network analysis problems was proposed. The problem of stance detection, an online social network analysis problem, was first tackled as an optimization problem. Furthermore, a new hybrid metaheuristic optimization algorithm was proposed for the first time in the current study, and the algorithm was compared with various methods. The analysis of the findings obtained with accuracy, precision, recall, and F-measure classification metrics demonstrated that our method performed better than other methods.     

地矿三维建模及可视化的研究

地矿三维建模及其三维可视化对地质、矿藏信息的获取及地质现象解释具有十分重要的意义。采用基于TIN的多层DEM方法，实现层状地矿实体模型的三维重构，借助于OpenGL图形库实现了层状实体模型的三维可视化，并对模型进行了光照渲染、漫游和切割等操作。