我国就业结构变动与技术升级研究*

本文利用一个全国代表性数据库，计算了2003—2013年间城镇劳动力市场就业结构变动情况，并从人口学视角进行了变动原因的分解分析。本文发现过去十年我国技术进步并没有带来显著的劳动参与率下降。相反，除了常规操作性工作，其他三类就业占比都在增长。增长原因既有人口结构变化带来的“结构效应”，也有需求变动带来的“倾向效应”。从绝对值看，由于目前我国常规性就业占比仍然远大于美国，这意味着如果人工智能等技术大力推进，仍然可能对我国劳动力市场带来更大冲击。因此，笔者也参照已有文献做法估计了我国就业中可能被替代的潜在数量和比例，以及未来行业间就业转换的方向。基于这些计算和讨论，本文对当前技术进步采纳及就业应对提出了若干政策建议。     

Enterprise knowledge graphs with applications to continuous monitoring systems

Continuous monitoring systems generate continuous business intelligence by gathering text from a range of sources and typically storing that text information using snippets of text. Further, continuous monitoring systems derive information from those snippets, such as sentiment and new concepts emerging in the text, beyond the existing ontology and dictionaries. However, it is difficult for users to gather much additional information from large quantities of text snippets. As a result, this paper investigates introducing the notion of “knowledge graphs” into such systems and analyzes some of the potential applications and capabilities in business intelligence settings. As part of that analysis, this paper extends current-generation continuous monitoring systems to include knowledge graphs.     

海关行政执法自由裁量权人工智能辅助决策初探

笔者试图引入人工智能对执法中的自由裁量进行辅助决策,探索以科技手段减轻行政执法人员的执法难度,提升化解执法风险和廉政风险的能力。     

我国东中西部人工智能产业政策及发展对比研究

随着人工智能被提升为国家发展战略,各地方政府已经开始多方布局人工智能应用及其产业化发展。首先,对比分析了我国部分省市人工智能政策制定及产业发展的现状。然后,结合东、中、西部不同区域当前人工智能产业的发展水平,提出了我国东、中、西部人工智能产业发展存在的问题。最后,从政策制定、平台建设、产业驱动、产业协调和人才创新等方面提出我国人工智能产业区域均衡发展的建议,以期实现我国人工智能产业的良性有序发展。     

Commercial Mortgage Default: A Comparison of Logit with Radial Basis Function Networks

This article explores the use of artificial neural networks in the modeling of foreclosure of commercial mortgages. The study employs a large set of individual loan histories previously used in the literature of proportional hazard models on loan default. Radial basis function networks are trained (estimated) using the same input variables as those used in the logistic. The objective is to demonstrate the use of networks in forecasting mortgage default and to compare their performance with that of the logistic benchmark in terms of prediction accuracy. Neural networks are shown to be superior to the logistic in terms of discriminating between good and bad loans. The study performs sensitivity analysis on the average loan and offers suggestions on further improving prediction of defaulting loans.     

AI智能技术在服饰中的研究与应用

本文从人工智能的发展历程、涉及领域、优势等角度出发,浅析了人工智能在服饰中的运用,以及得出了人工智能的价值在服装界得到了高度重视。逐步分析且展现出这项技术在服装中展现的魅力,其中包括:服装面料、服装设计、服装制版、服装裁剪、服装工艺、服装后整理和服装销售,整个服装产业链几乎都有AI技术的身影。我们知道,AI这项技术出现的初衷都是服务于人类,那么在服装上,我们确实也是需要这样的一项工具来提高我们产业的整体效率,越来越多企业正在加强这项技术在服装上的研发,同时已有的产品也得到了越来越多人们好的青睐。     

V-ROOM: a virtual meeting system with intelligent structured summarisation

With the growth of virtual organisations and multinational companies, virtual collaboration tasks are becoming more important for employees. This paper describes the development of a virtual meeting system called V-ROOM. An exploration of facilities required in such a system has been conducted. The findings highlighted that intelligent systems are needed, especially since information that individuals have to know and process is vast. The survey results showed that meeting summarisation is one of the most important new features that should be added to virtual meeting systems for enterprises. This paper highlights the innovative methods employed in V-ROOM to produce relevant meeting summaries. V-ROOM’s approach is compared to other methods from the literature, and it is shown how the use of metadata provided by parts of the V-ROOM system can improve the quality of summaries produced.     

基于半监督聚类的网络嵌入方法

GEMSEC(graph embedding with self clustering)在计算节点特征的同时学习节点聚类,通过强制将节点进行聚类来揭露网络中的社区结构,但未考虑类别标签信息,导致学到的节点嵌入缺乏区分性。针对这一问题,提出了一种基于半监督聚类的网络嵌入方法(NESSC),将随机游走序列和少量节点类别标签作为输入,在计算节点特征和学习节点k-means聚类的过程中,利用类别标签信息指导聚类过程,同时重构已知节点类别标签信息,学习具有区分性的节点表示。在6个真实网络上进行节点聚类和节点分类评测实验,实验结果显示,NESSC方法明显优于无监督网络嵌入方法DeepWalk和GEMSEC,可以通过加入节点的标签信息来提高网络嵌入的效果。因此,通过网络节点的嵌入,可以高效地提取网络的有用信息,对于相关网络嵌入研究具有一定的参考价值。     

A Flexible Neuro‐Fuzzy Approach for Improvement of Seasonal Housing Price Estimation in Uncertain and Non‐Linear Environments

Changes in housing price affect both individuals and government since they have substantial influence on the socio‐economic conditions. Valuations of housing are necessary in order to assess the benefits and liabilities in housing sector. This study presents a flexible meta‐modelling approach for improvement of housing price estimation in ambiguous and complex environments. It is composed of artificial neural network (ANN) and fuzzy linear regression (FLR). Seven FLR models are considered to cover latest approaches and viewpoints. Also, ANN is applied to data sets. The preferred FLR model is selected via mean absolute percentage of error (MAPE) for further considerations, and then the preferred FLR model and the best structure of ANN are applied to the data set. Finally, the preferred model is selected based on MAPE. The intelligent approach of this study is applied for estimation and forecasting housing price in Iran. The housing price in Iran mainly is based on eight economic indices including currency, oil income, general index, house service pricing index, rate of informal market, gross domestic production in basic price, added value of oil group and construction materials price. FLR is identified as the preferred model with lowest MAPE for housing price forecasting in Iran. This shows that the housing market of Iran is associated with severe environmental fuzziness and ambiguity. This is the first study that introduces a flexible neuro‐fuzzy approach for improved estimation and forecasting of housing price in noisy, complex and uncertain environments.