Semi-supervised model adaptation for statistical machine translation

Statistical machine translation systems are usually trained on large amounts of bilingual text (used to learn a translation model), and also large amounts of monolingual text in the target language (used to train a language model). In this article we explore the use of semi-supervised model adaptation methods for the effective use of monolingual data from the source language in order to improve translation quality. We propose several algorithms with this aim, and present the strengths and weaknesses of each one. We present detailed experimental evaluations on the French–English EuroParl data set and on data from the NIST Chinese–English large-data track. We show a significant improvement in translation quality on both tasks.     

CONTEXTUALIZING LEARNING OBJECTS USING ONTOLOGIES

Educational research over the past three years has intensified such that the context of learning resources needs to be properly modeled. Many researchers have described and even mandated the use of ontologies in the research being conducted, yet the process of actually connecting one or more ontologies to a learning object has not been extensively discussed. This paper describes a practical model for associating multiple ontologies with learning objects while making full use of the IEEE LOM specification. The model categorizes these ontologies according to five major categories of context based on the most popular fields of study actively being pursued by the educational research community: Thematic context, Pedagogical context, Learner context, Organizational context, and Historical/Statistical context.     

Automatic coding of dialogue acts in collaboration protocols

Although protocol analysis can be an important tool for researchers to investigate the process of collaboration and communication, the use of this method of analysis can be time consuming. Hence, an automatic coding procedure for coding dialogue acts was developed. This procedure helps to determine the communicative function of messages in online discussions by recognizing discourse markers and cue phrases in the utterances. Five main communicative functions are distinguished: argumentative, responsive, informative, elicitative, and imperative. A total of 29 different dialogue acts are specified and recognized automatically in collaboration protocols. The reliability of the automatic coding procedure was determined by comparing automatically coded dialogue acts to hand-coded dialogue acts by a human rater. The validity of the automatic coding procedure was examined using three different types of analyses. First, an examination of group differences was used (dialogue acts used by female versus male students). Ideally, the coding procedure should be able to distinguish between groups who are likely to communicate differently. Second, to examine the validity of the automatic coding procedure through examination of experimental intervention, the results of the automatic coding procedure of students, with access to a tool that visualizes the degree of participation of each student, were compared to students who did not have access to this tool. Finally, the validity of the automatic coding procedure of dialogue acts was examined using correlation analyses. Results of the automatic coding procedure of dialogue acts of utterances (form) were related to results of a manual coding procedure of the collaborative activities to which the utterances refer (content). The analyses presented in this paper indicate promising results concerning the reliability and validity of the automatic coding procedure for dialogue acts. However, limitations of the procedure were also found and discussed.     

Modern Teaching and Training in Metallurgical Engineering

This contribution deals with advanced educational technologies needed to equip customers from higher education institutions, research and industry with efficient tools supporting their work and operating new skills‐training methods. The challenges are reducing the training costs, improving quality and increasing the number of graduates in engineering departments. The concept of a Virtual Lab based on the combination of various teaching methods and tools is presented. Principles of mathematisation in metallurgical education and training are discussed. An example of online course designed in the form of Virtual Lab is demonstrated.     

Implicit, long-term spatial contextual memory.

Learning and memory of novel spatial configurations aids behaviors such as visual search through an implicit process called contextual cuing (M. M. Chun & Y. Jiang, 1998). The present study provides rigorous tests of the implicit nature of contextual cuing. Experiment 1 used a recognition test that closely matched the learning task, confirming that memory traces of predictive spatial context were not accessible to conscious retrieval. Experiment 2 gave explicit instructions to encode visual context during learning, but learning was not improved and conscious memory remained undetectable. Experiment 3 illustrates that memory traces for spatial context may persist for at least 1 week, suggesting a long-term component of contextual cuing. These experiments indicate that the learning and memory of spatial context in the contextual cuing task are indeed implicit. The results have implications for understanding the neural substrate of spatial contextual learning, which may depend on an intact medial temporal lobe system that includes the hippocampus (M. M. Chun & E. A. Phelps, 1999). (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

From children's hands to adults' ears: Gesture's role in the learning process.

Children can express thoughts in gesture that they do not express in speech--they produce gesture-speech mismatches. Moreover, children who produce mismatches on a given task are particularly ready to learn that task. Gesture, then, is a tool that researchers can use to predict who will profit from instruction. But is gesture also useful to adults who must decide how to instruct a particular child? We asked 8 adults to instruct 38 third- and fourth-grade children individually in a math problem. We found that the adults offered more variable instruction to children who produced mismatches than to children who produced no mismatches--more different types of instructional strategies and more instructions that contained two different strategies, one in speech and the other in gesture. The children thus appeared to be shaping their own learning environments just by moving their hands. Gesture not only reflects a child's understanding but can play a role in eliciting input that could shape that understanding. As such, it may be part of the mechanism of cognitive change. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

大庆长垣以东古隆起天然气成藏条件的差异性

由长垣以东地区深层天然气分布特征，指出安达—肇州和朝阳沟—肇东古隆起是天然气运移指向和聚集的有利场所。造成安达—肇州古隆起上天然气显示好于朝阳沟—肇东古隆起上天然气显示，前者已获得了大量工业气流，而后者仅见少量气流或显示。但由于天然气在供气条件、储集条件、盖层条件、圈闭条件和运移条件上的差异，造成安达—肇州古隆起本身天然气显示也存在差异，昌德构造、卫星—升平构造和汪家屯构造天然气显示好于模范屯—肇州西构造天然气显示。     

中国区域地质调查现代化与加强深部调查

从对区域地质调查的任务出发，总结了过去区域地质调查工作的特点，即基本上是沿用传统方式进行的。提出和论述了21世纪中国区域地质调查应有新发展，以适应战略转变的要求，即除去运输、通讯、记录及数字化成图等之外，更应当面向土地资源、矿产资源、地质灾害以及提高环境质量的需要, 用现代地学理论指导进行多学科的调查和采集、整理资料，实现3维多参数填图和理论概括，是关系现代地质科学发展的、具有深远战略意义的事。几十年的调查工作已证明深部的地质构造情况与地表层的观察结果是不一样的，与依据地表地质构造情况推测的深层情况也很不同。今后应加强深部探测，发展深部探测技术，取得不同深度层次的高分辨率成像图案，使人们对地球的认识提高到一个新的高度，形成新的空间—时间观念，以更有效地解决我国及地球的资源环境问题。     

It's a sign of the kind: Gestures and words guide infants' inductive inferences.

The role of words and gestures in guiding infants' inductive inferences about nonobvious properties was examined. One hundred seventy-two 14-month-olds and 22-month-olds were presented with novel target objects followed by test objects that varied in similarity to the target. Objects were introduced with a novel word or a novel gesture or with no label. When target and test objects were highly similar in shape, both 14- and 22-month-olds inferred that these objects shared a nonobvious property, regardless of whether the objects were labeled with a word or a gesture or with no label. When objects were labeled with the same word, both 14- and 22-month-olds generalized the nonobvious properties to objects that shared minimal perceptual similarity. Finally, when objects were labeled with the same gesture, 14-month-olds, but not 22-month-olds, generalized the nonobvious properties to objects that shared minimal perceptual similarity. These results indicate that 14-month-olds possess a more generalized symbolic system as they will rely on both words and gestures to guide their inferences. By 22-months of age, infants treat words as a privileged referential form when making inductive inferences. (PsycINFO Database Record (c) 2010 APA, all rights reserved)