校园户外坐具的语义诉求

从产品语义学的角度出发,阐述了产品外延性语义和内涵性语义的概念及关系,以校园户外坐具为线索,着重分析了在校园特定的语境下所关联的符号要素,并将其融入到户外坐具的内涵语义设计中,指出校园户外坐具在满足基本使用功能的外延语义传达基础上,更成为凸显校园特质和文化气息的内涵语义的传达载体,点明了校园户外坐具在语义传达上所应具备的多层次性。     

汉语分词中上文和下文重要性比较

上下文是统计语言学中获取语言知识和解决自然语言处理中多种实际应用问题必须依靠的资源和基础。近年来基于字的词位标注的方法极大地提高了汉语分词的性能,该方法将汉语分词转化为字的词位标注问题,当前字的词位标注需要借助于该字的上下文来确定。为克服仅凭主观经验给出猜测结果的不足,采用四词位标注集,使用条件随机场模型研究了词位标注汉语分词中上文和下文对分词性能的贡献情况,在国际汉语分词评测Bakeoff2005的PKU和MSRA两种语料上进行了封闭测试,采用分别表征上文和下文的特征模板集进行了对比实验,结果表明,下文对分词性能的贡献比上文的贡献高出13个百分点以上。     

基于上下文音节的维吾尔语语音合成系统研究

提出上下文音节为拼接单元,生成自然语音的一种维吾尔语语音合成方法。上下文音节有效地整合了邻近音节内外的音联现象,可以避免大部分协同发音影响。采用基于时域平滑技术对合成语音进行平滑处理,减少在拼接点处的语音失真和突变,从而提高了合成语音的自然度。     

The relative relativity of material and experiential purchases.

When it comes to spending disposable income, experiential purchases tend to make people happier than material purchases (Van Boven & Gilovich, 2003). But why are experiences more satisfying? We propose that the evaluation of experiences tends to be less comparative than that of material possessions, such that potentially invidious comparisons have less impact on satisfaction with experiences than with material possessions. Support for this contention was obtained in 8 studies. We found that participants were less satisfied with their material purchases because they were more likely to ruminate about unchosen options (Study 1); that participants tended to maximize when selecting material goods and satisfice when selecting experiences (Study 2); that participants examined unchosen material purchases more than unchosen experiential purchases (Study 3); and that, relative to experiences, participants’ satisfaction with their material possessions was undermined more by comparisons to other available options (Studies 4 and 5A), to the same option at a different price (Studies 5B and 6), and to the purchases of other individuals (Study 5C). Our results suggest that experiential purchase decisions are easier to make and more conducive to well-being. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Using context to resolve temporal ambiguity.

Three conditioned lick suppression experiments with rats examined the role of the context in the selection and integration of independently acquired interval relationships. In Experiment 1, rats were exposed to separate conditioned stimuli 1 and 2 (CS1–CS2) pairings with 2 different interval relationships, each in its own distinctive context, X or Y. The resultant integration was determined by the training context (X or Y) in which unconditioned stimulus (US)–CS2 backward pairings occurred, as assessed in a third neutral context (Z). In Experiment 2, rats experienced CS1–CS2 pairings with 2 different interval relationships as in Experiment 1, and then received US–CS2 pairings in both contexts X and Y. The testing context (i.e., X or Y) determined the resultant integration. In Experiment 3, rats were exposed to CS1–CS2 pairings in 2 different interval relationships each in different phases (i.e., Phases 1 and 2), and then in Phase 3 received US–CS2 pairings. The temporal context of testing (i.e., short or long retention interval) determined the resultant integration. Thus, both physical and temporal context can be used to disambiguate conflicting temporal information. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Social context and perceived effects of drugs on sexual behavior among individuals who use both heroin and cocaine.

Researchers have identified the association between the use of cocaine and sexual behavior as an important risk factor for HIV infection and have attempted to elucidate the nature of this association. Several lines of research have suggested that facilitation of sexual behavior during intoxication with cocaine may be because of the direct pharmacological effects of the drug (e.g., increase in sexual desire), whereas others have pointed to the importance of factors related to the context of drug use (e.g., opportunities for sexual behavior, expectations about the effects of the drug, social norms). The present study explored the perceived effects of cocaine and heroin on sexual behavior, as well as the social context of drug use as a function of drug type (cocaine vs. heroin), among 46 inner-city drug users who reported a history of regular use of both crack cocaine and heroin. Results indicated that compared to heroin, cocaine had deleterious effects on participants' perceived sexual desire and performance. Despite such deleterious effects on sexual behavior, cocaine was more frequently used with an intimate partner than heroin. Furthermore, participants did not differ in the extent to which they used the two drugs in other social contexts (e.g., with friends, family, or neighbors). These preliminary results suggest that the relationship between cocaine and sexual behavior, especially among long-term cocaine users, may be facilitated by opportunities for sex that exist in the context of cocaine use, rather than by the pharmacological effects of the drug. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

面向城市应急事件管理的监控视频智能解析

随着城市现代化进程的不断加速,建立高效、完善的城市应急事件管理系统迫在眉睫。遍布城市各处的监控摄像头拍摄的监控视频是城市应急事件管理系统重要的数据来源,但在现阶段主要依赖人工识别,费时费力且效果不佳。提出一种面向城市应急管理的监控视频智能解析方法,利用计算机视觉库获取视频中监控对象的数量和位置,生成其运动轨迹并匹配预设的异常运动模板,从而构建应急事件的规范化情境模型用以分析和预警。以某大学校园开放区域和某高新科技园区的监控视频为例,验证了所提出方法的可行性和合理性。     

Rank-Order Correlation-Based Feature Vector Context Transformation for Learning to Rank for Information Retrieval

As a crucial task in information retrieval, ranking defines the preferential order among the retrieved documents for a given query. Supervised learning has recently been dedicated to automatically learning ranking models by incorporating various models into one effective model. This paper proposes a novel supervised learning method, in which instances are represented as bags of contexts of features, instead of bags of features. The method applies rank-order correlations to measure the correlation relationships between features. The feature vectors of instances, i.e., the 1st-order raw feature vectors, are then mapped into the feature correlation space via projection to derive the context-level feature vectors, i.e., the 2nd-order context feature vectors. As for ranking model learning, Ranking SVM is employed with the 2nd-order context feature vectors as the input. The proposed method is evaluated using the LETOR benchmark datasets and is found to perform well with competitive results. The results suggest that the learning method benefits from the rank-order-correlation-based feature vector context transformation.     

基于社会化媒体情境的多维协同智能推荐

针对传统协同智能推荐技术的冷启动、数据稀缺性问题,为提高推荐算法的效率和准确性,提出一种基于社会化媒体情境的多维智能推荐算法模型。该模型将目标用户的属性特征、行为特征考虑到社会化媒体情境信息中,并动态实时捕捉用户在不同社会化媒体情境下的偏好倾向,利用联机分析处理(OLAP)技术对多维数据进行处理。该模型将用户间的社会化关系和所处的政治经济环境视为衡量用户相似的重要指标,同时使用皮尔森系数和云模型来计算用户间各特征的相似度,并以此为推荐基础向用户呈现更个性化和定制化的推荐结果。实验结果表明,该模型的推荐结果的平均绝对误差明显小于传统的协同智能推荐和单纯的基于云模型推荐技术。     

基于时间上下文跟踪学习检测的指尖跟踪方法

针对在基于视频的空中签名认证系统中,现有方法无法满足指尖跟踪的准确性、实时性和鲁棒性要求的问题,在对比研究目前常用的多种跟踪方法的基础上,提出一种基于时间上下文的跟踪-学习-检测(TLD)方法。在原始TLD算法的基础上引入时间上下文信息,即相邻两帧间指尖运动具有连续性的先验知识,自适应地缩小检测和跟踪的搜索范围,以提高跟踪的速度。对12组公开的1组自录的视频序列的实验结果表明,改进后的TLD算法能够准确地跟踪指尖,并且跟踪速度达到43帧/秒;与原始TLD跟踪算法相比,准确率提高了15%,跟踪速度至少提高1倍,达到了指尖跟踪的准确性、实时性和鲁棒性要求。