incremental learning of triadic plsa for collaborative filtering

Beijing University of Technology (BJUT); Beijing Municipal Lab of Brain Informatics; Chinese Society of Radiology; National Natural Science Foundation of China (NSFC); State Administration of Foreign Experts Affairs

An application of the analytic hierarchy process and fuzzy logic inference in a decision support system for forage selection

Forage selection plays a prominent role in the process of returning cultivated lands back into grasslands. The conventional method of selecting forage species can only provide attempts for problem-solving without considering the relationships among the decision factors globally. Therefore, this study is dedicated to developing a decision support system to help farmers correctly select suitable forage species for the target sites. After collecting data through a field study, we developed this decision support system. It consists of three steps: (1) the analytic hierarchy process (AHP), (2) weights determination, and (3) decision making. In the first step, six factors influencing forage growth were selected by reviewing the related references and by interviewing experts. Then a fuzzy matrix was devised to determine the weight of each factor in the second step. Finally, a gradual alternative decision support system was created to help farmers choose suitable forage species for their lands in the third step. The results showed that the AHP and fuzzy logic are useful for forage selection decision making, and the proposed system can provide accurate results in a certain area (Gansu Province) of China.

离群模糊核聚类算法

一般说来,离群点是远离其他数据点的数据,但很可能包含着极其重要的信息.提出了一种新的离群模糊核聚类算法来发现样本集中的离群点.通过Mercer核把原来的数据空间映射到特征空间,并为特征空间的每个向量分配一个动态权值,在经典的FCM模糊聚类算法的基础上得到了一个特征空间内的全新的聚类目标函数,通过对目标函数的优化,最终得到了各个数据的权值,根据权值的大小标识出样本集中的离群点.仿真实验的结果表明了该离群模糊核聚类算法的可行性和有效性.

知识表达语言(NFA)及其在石油测井解释专家系统中的应用

石油测井解释是一项逻辑推理和数值计算交错进行的复杂过程。为了描述测井解释专家的这种知识、经验并模拟其思维方式,在扩充纯产生式规则的基础上,我们开发了知识表达语言——NFA,它把逻辑推理和数值计算综合成统一的形式。石油测井解释专家系统 LIX 先后在 INTERDATA-85机和 PE-3230机上实现,现场(胜利油田)运行近两年,解释了130余口井,符合率94%以上。LIX 实质上是 NFA 语言的解释系统,它的研制成功,说明了 NFA 语言的有效性和实用性。

基于数字化生产模型的在线故障诊断技术研究

提出了一种基于数字化的生产模型,使用控制图、故障树分析和专家知识,能够进行制造过程实时监控的诊断,该模型提高了故障诊断系统的可靠性,并提供了可实际操作的可视化建模工具。所开发的在线统计过程控制系统能够根据生产事件的监测,动态响应制造过程变化。该系统运用可视化建模工具,根据专家经验进行故障树建模,通过故障树自动生成专家系统诊断规则库,实现诊断知识的自动获取。将该系统应用于汽车变速箱装配过程的检测与故障诊断,验证了方法的有效性。

基于最小二乘拟合的模糊隶属函数构建方法

针对当前模糊隶属函数构造方法中存在的问题,提出一种构造模糊隶属函数方法.采用最小二乘法拟合离散数据来获得隶属函数.为减小拟合误差,采用了3项措施以达到预期目标.所构建的隶属函数,对任意输入物理量可直接得到其对应模糊语言变量的隶属度,从而有效避免专家指定隶属度的主观臆断性及不一致性.该方法简单、求解精度高,具有广泛适用性和较强的应用价值.仿真结果证实了该方法的有效性.

基于模糊统计的模糊Petri网token确定方法

在模糊Petri网应用研究中,普遍存在模糊token由专家直接给出或主观假定的问题。基于这种情况,提出了通过模糊统计法来获得库所的模糊token,为成功应用模糊Petri网理论创造了条件。给出了计算模糊token的通用形式化算法。实例论证了模糊统计法在求取模糊token时的可行性与有效性。

巴丹吉林沙漠及其南缘地区的水化学和环境同位素特征与环境演变

The Badain Jaran Desert lies on the Alashan Plateau in western Inner Mongolia. Because of huge dunes, permanent lakes and on the northern fringe of the Asian summer monsoon, the Badain Jaran Desert has been drawing attentions of many experts. And they have made great progress in dune’s geomorphology, botany in desert, paleoclimate change and other study areas. We analyzed environmental isotope and ion chemistry in lakes and groundwater of the desert and southeastern area, and collected some other evidences from 14C dating, fossils and archeology. According to chemical analysis, we discuss the difference spatial character of ion chemistry and environmental isotope in lakes and groundwater of the desert and adjacent. Contrasting with ion chemistry and isotope results in other arid area, we argue origin of groundwater and lakes in the desert area, and get a preliminary understanding of desert lakes’ evolution during Holocene. Some main conclusions were drawn as follows: 1. It has a obvious difference in hydrophysical parameters between lakes and groundwater in the desert and margin. 2. The results of ion analysis show that Na+ and Cl- are dominant in most lakes of the desert. Meanwhile, Na+ 、Cl- and HCO3- are dominant in groundwater of the desert and adjacent, and alsoMg2+、Ca2+、and NO3- have more percentage than in lakes. 3. Owing to different solubilities, the conten of main ions in water varies with the content of TDS. Whereas the content of TDS is over 100 g/L, the content of SO42-、HCO3-、Mg2+and Ca2+ in lakes descend. 4. The result of isotope analyzing indicate the lakes and groundwater in southeast desert have a similar vaporing trend with the groundwater in the southeast margin of the desert. It imply there would have some kind of contact between groundwater in margin and lakes of southeast desert. 5. Contrasting with isotope results of groundwater in other arid area, it show that the groundwater in the desert and Yabulai area should be phreatic water which have a low water table. Therefore, we conclude that the groundwater in southeast part of the desert and southern margin mainly are recharged by precipitation of local abundant rainfall and groundwater of low mountain of southern area. 6. And all of these evidences, which are different from salinity, the content of CO32- and geological data, show that the bigger northern lake group and southeastern lake group in the desert have different groundwater replenishing system because a fold belt lie between of the two group lakes and obstruct them in landform. and HCO3- 7. The 14C dating results of fossil and lacustrine deposits show that there maybe have a wider range of shoreline during early and middle Holocene than today. 8. By the discovery and study of some pieces of pottery and fine stoneware, we preliminary conclude that there maybe have some certain amount of early human activities in the Badain Jaran Desert.

活动构造区区域地壳稳定性研究——以金沙江水电梯级工程规划区为例

The study of regional crustal stability of active tectonic region basically includes analysis of recent activity of Earth's crust, single factor assessment, study of complexity, and comprehensive assessment of crust stability. In this thesis, some work are made as follows: · Based on abundant data from gravity field, aeromagnetic survey, magnetism, magnetotelluric deep sounding, remote sensing and geotectonic as well as earthquakes observed in recent years around this region and adjacent zones, we can get a through understanding about the structural features and activity of the earth's crust in Chuan-Dian region. The results from explosion earthquake and telluric electromagnetic sounding are consistent with the structural features of the crust manifested by the geophysical field. The data of deep geologic structures are important for us to work out a vivid three-dimensional model of the earth's crustal structure of the Jinsha River region. According to a synthesis, the author of this thesis proposes some indicators for dividing the faulted blocks. It can also be inferred that the movement of the Chuan-Dian faulted block, which is the relatively active part of southwestern China, is controlled by the boundary faults, and the intensive activity and deformation are concentrated along the boundaries of the block. · Mainly discussing respectively the mechanism and laws of active faults, earthquakes, and geological hazards activity, and their influences on the stability and security of engineering, also trying to probe into the way to assess the risk of single factor in this section. Especially with the method of fractal geometry, the thesis has discussed how to study the complexity of each factor. These geologic hazards which are distributed at the uppermost part of the crust in this region form a typical mountainous set of the active tectonic areas. The results of survey show that some slopes are liable- to -sliding with a weak layer of low shear strength. Occurrences of landslides are to a great extent related to local geological structures, in particular active faults. This is why numerous landslides have occurred simultaneously around the epicenter of a strong earthquake or the center of a strong rainfall, which are related to active faults. · The analysis of the crustal stability is based on a regional grid division, and a fuzzy comprehensive analysis method is used to determine the grade of the quality in each grid. The evaluation factors and their weights are taken from the results of the hierarchical analysis. The evaluation indexes consist of qualitative and quantitative ones. The qualitative ones can be quantified through the experts weighing system, while the quantitative ones can be obtained from statistical analysis. For quality grades, four levels are used: stable, essentially stable, sub-stable, and unstable. The results of the evaluation on Jinshajiang region demonstrate that the crustal stability become distinctly worse in the areas controlled by active deep faults. Therefore, detailed investigations on the active faulting and geologic hazards, include earthquake activity are especially necessary for those areas adjacent to the deep fault belts. On the bases of the data available and the survey results, we have made a preliminary assessment for the construction conditions and adaptability of every planned site in the middle or lower reaches of Jinsha River. Finally, the thesis prospected the vista of the study of crustal stability.

面向决策支持的综合地质信息系统功能开发和GIS应用

Synthetic Geology Information System(SGIS) is a part of the theory of Engineering Geomechanics-mate-Synthetics(EGMS), is also a development of its technical methodology. SGIS includes ways of geology engineering investigation, design, and construction. Although SGIS has an integrate theory frame, and some parts of it have gained great progress, the completion of SGIS is a continuous and accumulative process. This paper analyses the ways and principle of building knowledge database and model database, summarizes the experts' experience on exploration methods selection and the characters of exploration models, combining with the application of Decision Support System(DSS) in Decision support of Synthetic Exploration Methods for Railway engineering Geology. By the analysis of hierarchy structure of the model database, the effects of geology engineering factors on the selection of exploration methods are expressed. By the usage of fuzzy patterns recognize, hierarchy structure analysis, fuzzy collection closement analysis etc, the software of DSS for engineering design and construction are developed. At same time, by the development of Monitoring Data Analysis System and experiment data management system of Hydro-power project, this paper discussed the data management of science experiment of Hydro-power project by the usage of synthetic database and the usage of Geography Information System(GIS) and DSS technics. The technic of visual operation of data process and project monitoring system are presented. The intelligence algorithm of self-adoption is carried out to improve the data process and analysis of monitoring. Items of the project theoretical analysis and data process are designed in detail. All the theory and technical methods presented in this paper are one part of SGIS, in which the application of DSS and GIS, is an important step of the progress and completion of SGIS.

态度结构一致性对态度-行为关系的影响：个体与群体水平的研究

Social psychologists have long been interested in understanding the conditions under which attitudes influence behaviors, and they have formed two relatively independent theoretical perspectives in the attitude-behavior domain. One perspective, which we adopt in the present paper, focuses on investigating the effect of attitude strength on the attitude-behavior relationship. Specifically, the present research investigates the role of structural consistency, a dimension of attitude strength, on the attitude-behavior relationship. Structural consistency has typically been defined as the extent to which one aspect of individual’s attitude is consistent with either his or her overall attitude toward the object or the other aspect of the attitude, including affective-cognitive consistency (ACC), evaluative-cognitive consistency (ECC) and evaluative-affective consistency (EAC). The ECC and EAC can be integrated into the attitude base. Despite the evidence that structural consistency is associated with attitude-behavior relationship, there are some limitations in previous studies: (1) attitude base and ACC have been rarely studied together ;( 2) researchers often used the term behavior in a broad sense to encompass behavioral intentions as well as actual behaviors ;( 3) the effects were all investigated at individual level. The present research tried to explore the effects of ACC and attitude base on attitude-behavioral intention-actual behavior relationships at individual and group levels in different behavior areas including shopping, supportive behavior towards officials, and offensive behavior towards governors and central and local governments. The roles of affective and cognitive components of attitudes on general attitudes and following behaviors were also examined in this research. The results indicated that: A. At individual and group levels, the effect of structural consistency on attitude –intention relation is as the same as its effect on attitude-behavior relation, but it varies in different areas. On the other hand, the effect of structural consistency on explicit attitude-behavior relation is different from its effect on implicit attitude-behavior relation. B. The cognitive components of the individuals and groups’ implicit attitudes towards officials, the implicit and explicit attitudes towards governors, and the public’s attitudes towards central and local governments exert the stronger influence on general attitudes. Similarly, the cognitive components also exert the stronger influence on groups’ supportive behaviors towards officials and offensive behaviors towards governors and governments. C. Both group attitude and emotion can influence the group offensive behavior towards governors. The group attitude exerts its influence through the mediating effects of behavior intention and group emotion. The group emotion which is the core factor can exert the direct influence on group offensive behavior towards governors. D. At individual and group levels, the implicit and explicit measures should be used together to collect the attitude data, because the implicit and explicit attitudes can both predict behaviors under most conditions. It is the best implicit attitude measure to use the experts’ evaluations of listed words. And the emotion should be measured by the experts’ evaluations of listed words if the emotion data is used to predict the following behavior.

工作幸福感的结构、测量及作用机制

With the emergence and development of positive psychology, happiness has been the focus of academia and business. However, there is no uniform measure of happiness, because of many different theories of happiness, which are not compatible with others. It bounds the further development of happiness theory. It is also the same with the research of work well-being, which refers to the emotional experience and quality of psychological functioning of employee in the workplace. Subjective well-being (SWB) and psychological well-being (PWB) are two major theories of happiness. Prior research has demonstrated the integration of these two theories theoretically, but still needs more empirical support. Besides, in line with the development of positive psychology, a body of knowledge about positive leadership is advocated. Transformational leadership is treated as one kind of positive leadership, since it emphasizes the leader’s motivational and elevating effect on followers. But the extent to which the transformational leadership can enhance work well-being, and what the mechanism is, these are the questions need to be explored. Based on the integration of SWB and PWB, this research tried to investigate the structure, measurement and mechanism of work well-being, and combining with the theory of transformational leadership, this study also tried to investigate the relationship between transformational leadership and work well-being. The structure and measurement of work well-being, the relationships between work well-being and job characteristics (including job resources and job demands), the relationships among transformational leadership, job resources, work well-being and corresponding outcomes, the relationships among transformational leadership, job demands, work well-being and corresponding outcomes, and the relationships among transformational leadership, group job characteristics, group work well-being and corresponding group outcomes were explored by using content analysis, Subject Matter Experts (SMEs) discussion, and structural questionnaire surveys. More than 7000 subjects were surveyed, and Explore Factor Analysis (EFA), Confirm Factor Analysis (CFA), Structural Equation Modeling (SEM), Hierarchical Linear Modeling (HLM) and other statistics methods were used. The following is the major conclusions. Firstly, work well-being is a two high-order factors structure, which includes affective well-being (AWB) and cognitive well-being (CWB). AWB is similar to SWB, and CWB is similar to PWB. Besides, the construct of AWB includes sub-dimensions of positive emotional experience and negative emotional experience. And the construct of CWB consists of work autonomy, personal growth, work competent, and work significance. Secondly, the relationships between job characteristics and AWB and CWB are different. On one hand job demands are directly related to AWB, and are indirectly related to CWB through the full mediation of AWB, on the other job resources are directly related to CWB, and are indirectly related to AWB through the full mediation of CWB, which means AWB and CWB reciprocally influences each other in the model of job demands-resources. These results were concluded as the process model of work well-being. Thirdly, AWB and CWB are positively related to many workplace outcomes, including job satisfaction, group satisfaction, organizational commitment, turnover intention, job performance, organizational citizenship behavior (OCB), and general psychological health and general physiological health. Fourthly, transformational leadership is indirectly related to CWB through the full mediation of job resources, and is related to AWB through the partial mediation of job demands. Meanwhile, transformational leadership is related to many workplace outcomes through the mediation of job characteristics and work well-being. These results implied that transformational leadership is indeed one kind of positive leadership. Fifthly, in the group level, transformational leadership is indirectly related to group CWB through the full mediation of group job resources, and is related to group AWB through the full mediation of group job demands. Group AWB has positive influence on group CWB, but not vice versa. Group job characteristics and group work well-being fully mediate the relationships between transformational leadership and intragroup cooperation and group performance.

管理干部胜任特征评价方法的初步研究

Competency Assessment Method (CAM) is an important technique of Human Resource Management and Development in theory and practice, especially in personnel selection and training. Based on literatures of related fields, the thesis explored the feasibility of CAM in China. The main results found in this study are as follows: 1. Competencies scored in Behavioral Event Interviews (BEI) are not influenced by length of protocol, by performance in the preceding year. Average level and maximal level of complexity correlate significantly with length of protocol. Total competency frequency of outstanding executives is not significantly different from that of typical executives. These results support McCleland's view. But there is significant correlation between length of protocol and competency frequencies, which which is not agreed by McCleland. We found that competency scores using coding standard of average level and maximal level of complexity show more reliability than that using coding standard of competency frequencies. But this isn't confirmed by McCleland. 2. Inter-rater reliability was studied. The results indicated: total Category Agreement (CA) is 55.45%, over 70 percent of 20 competencies of the inter-rater reliability coefficients based on the classical test theory are significantly correlated. G coefficient based on the generalization theory is 0.85697. 3. Study of criterion sample shows that manager's competencies of China's communication enterprise are as follows: Impact and Influence, Organization Commitment, Information Seeking, Achievement Orientation, Team Leadership, Interpersonal Understanding, Initiative, Market Awareness, Self-confidence, Developing Others. This result in similar to the generic competency model of managers presented in Spencer's book. 4. CAM showed more advantages than the method of experts panel judgement.

Learning World Models in Environments with Manifest Causal Structure

This thesis examines the problem of an autonomous agent learning a causal world model of its environment. Previous approaches to learning causal world models have concentrated on environments that are too "easy" (deterministic finite state machines) or too "hard" (containing much hidden state). We describe a new domain --- environments with manifest causal structure --- for learning. In such environments the agent has an abundance of perceptions of its environment. Specifically, it perceives almost all the relevant information it needs to understand the environment. Many environments of interest have manifest causal structure and we show that an agent can learn the manifest aspects of these environments quickly using straightforward learning techniques. We present a new algorithm to learn a rule-based causal world model from observations in the environment. The learning algorithm includes (1) a low level rule-learning algorithm that converges on a good set of specific rules, (2) a concept learning algorithm that learns concepts by finding completely correlated perceptions, and (3) an algorithm that learns general rules. In addition this thesis examines the problem of finding a good expert from a sequence of experts. Each expert has an "error rate"; we wish to find an expert with a low error rate. However, each expert's error rate and the distribution of error rates are unknown. A new expert-finding algorithm is presented and an upper bound on the expected error rate of the expert is derived.