英翻译成德，大家来帮忙找找茬。

orionsnow

没有放在英语学习的原因是，科技词太多，另外除了语法错误，我更希望发现意思表达方面的问题

另外就是我更想听应用学科，特别是生物学医药方面的回馈，毕竟顾客就是上帝

我已经请人做过基本语法检查了，请大家随便读读，然后看看那里意思表达还不到位。

$送花$

谢谢了，欢迎用任何方式找任何错误，包括标点错误，词性错误。 对正式发表的文章，我还是想尽量做到完美的。

changed

[ 本帖最后由 orionsnow 于 2008-12-8 15:25 编辑 ]

Wildblueten

帮顶。
先简单看了看英文。觉得单复数和冠词应该再check一下下。
我德语不灵光，没有回答楼主问题，不好意思地走了。。。

orionsnow

原帖由 Wildblueten 于 2008-10-2 10:44 发表
帮顶。
先简单看了看英文。觉得单复数和冠词应该再check一下下。
我德语不灵光，没有回答楼主问题，不好意思地走了。。。

英语语法错误也欢迎提阿，能不能具体点。

我的确是单复数还有定冠词方面需要小心。

刚才又请德国同事帮忙读了一遍，找到不少错误，学了不少新东西

Wildblueten

偶不是英语强人哦，就是觉得这里亲切，班门弄斧拉$害羞$

被杂志社的编辑和reviews折磨多了，拿点经验出来大家分享。大家互拍吧，共同进步。

Model-based \citep{Royston1999} and test-based \citep{Dosemeci1998}
approaches are most commonly used to analyze the (后面没有定语哦，这个the不一定要加吧，这个我不太确定) dose-response
relationship. In the problems（do you mean questions?） mentioned above, the null hypothesis,
which usually assumes no difference or no changes（单复数一致，这里ms单数合适）, should be tested
first. After the null hypothesis is rejected at a certain global level
$\alpha$, a model selection procedure is used to detect the
relationship. However, the test and model selection procedure
afterwards depend seriously on the dose-response shape, which is
unknown before the whole procedure. In this situation, a procedure
which has good power over the alternative space, is required. (好像少个逗号)An
order restricted likelihood ratio test is developed by
\cite{Robertson1988}. Other than the single test, a multiple
comparison procedure is developed by \cite{Mukerjee1987} who also
recommended orthogonal contrasts which have a simple power function.


The use of order-restricted hypotheses is a common approach which
increases power by narrowing the alternative space. Hereby,
Change-point, Simple-order and Simple-tree are three common types of
order restrictions. Several approaches are available for these
problems, such as max-t statistics according to \cite{Hirotsu2000},
which can be formulated as maximum contrast approach belonging to
the broader class of Multiple Contrast tests (MCT). The disadvantage
of MCT is that it makes model selection after the global null
hypotheses is rejected. The local finding rate of the true model of
the alternative is low. Recently, \cite{Anraku1999}, \cite{Zhao2002}
and \cite{Ninomiya2005} developed information criterion based（有点歧义哦？which based on? or ...）
log-likelihood method for model selection approaches under certain
types of order restriction. These methods have better finding rate
of the true model, but have no FWER-control. They treat the null
model as one of the possible model among all others. These methods
are not constructed as hypothesis tests to reject the alternatives.
We will compare these two methods in Change-point detection problem
and other order restriction problem(other order不是一个专署名词吧？如果不是，应该是the other...problem/other problems). Also we will present a modified
information criterion which can control the FWER-control for model
selection under certain order restriction(s).(或者 a certain order restriction)

[ 本帖最后由 Wildblueten 于 2008-10-2 14:22 编辑 ]

orionsnow

原帖由 Wildblueten 于 2008-10-2 14:20 发表
偶不是英语强人哦，就是觉得这里亲切，班门弄斧拉$害羞$

被杂志社的编辑和reviews折磨多了，拿点经验出来大家分享。大家互拍吧，共同进步。

Model-based \citep{Royston1999} and test-based \citep{Dosemeci ...

$送花$


谢谢了，刚才德文部分也请德国同事修改过了， 我先在把修改稿贴在下边。打算论文也写中文摘要，等导师同意了，把中文发下就更好对照了。

另外看来购买电子纠正软件对我来说挺必要的了， 一般的单复数，还有从句问体，用软件查起来很方便，况且，我的整个论文估计都需要做以上检查


\section*{Zusammenfassung}


In den Feldern der Biologie, Epidemiologie und Pharmakologie, wie gefunden ist, dass ein passendes Modell der Dosis-Wirkungs-Beziehung unter mehreren Gruppen eine allgemeine Frage beschreibt. Zum Beispiel werden Fragen, wie "Leiden Arbeiter der Nickel-Raffinerie mit größerer Wahrscheinlichkeit Lungenkrebs als andere?", "Wie bewirken Alter der Eltern die spontane Abtreibungsrate?" "Ist dort irgendeine Struktur-Änderung zwischen Motiven und 'junk-DNA' auf den DNA-Folgen?" und "Sollten wir Standarde für die Feinstaub in der Luft festlegen? Wenn die Antwort Ja, dann wie viel ist?" werden oft gefragt.

Modell-basierte und Test-basierte Methoden werden meistens zur Analyse der Dosis-Wirkungs-Beziehung verwendet. Für die Problem, die oben erwähnt sind, sollte die Nullhypothese, die gewöhnlich keinen Unterschied oder keine Änderungen annimmt, zuerst geprüft werden. Nachdem die Nullhypothese am bestimmten globalen $ \alpha$ zurückgewiesen ist, wird ein Musterauswahl-Verfahren verwendet, um die Beziehung zu entdecken. Jedoch hängen der Test und das Musterauswahl-Verfahren später ernstlich von der Gestalt der Dosis-Wirkungs ab, die vor dem ganzen Verfahren unbekannt ist. In dieser Situation ist ein Verfahren, das gute power über den alternativen Raum hat, erforderlich. Ein Eingeschränkter Wahrscheinlichkeitsverhältnis-Test einer Ordnung wurd durch \cite {Robertson1988} entwickelt. Ander als der einzelne Test wurd ein vielfaches Vergleich-Verfahren durch \cite {Mukerjee1987} entwickelt, der auch orthogonale Unähnlichkeiten empfahl, die eine einfache Potenzfunktion haben.


Der Gebrauch von Ordnungseingeschränkten Hypothesen ist eine einheitliche Methode, die power vergrößert, den alternativen Raum einengend. Hiermit sind Änderungspunkt, Einfache Ordnung und Einfacher Baum drei allgemeine Typen von Ordnungsbeschränkungen. Mehrere Annäherungen sind für diese Probleme, wie Max-T-Statistik gemäß \cite {Hirotsu2000} verfügbar, der als maximale Kontrastannäherung formuliert werden kann, die der breiteren Klasse von multiplen Kontrasttests (MKT) gehört. Der Nachteil von MKT ist, dass er Musterauswahl power, nachdem die globalen Nullhypothesen zurückgewiesen werden. Die lokale Entdeckungsrate des wahren Modells der Alternative ist niedrig. Kürzlich stützte \cite {Anraku1999}, \cite {Zhao2002} und \cite {Ninomiya2005} entwickeltes Informationskriterium Methode der Klotz-Wahrscheinlichkeit für Musterauswahl-Annäherungen unter bestimmten Typen der Ordnungsbeschränkung. Diese Methoden haben eine bessere Entdeckungsrate des wahren Modells, aber gibts keine FWER-Kontrolle. Sie behandeln die Nullhypothese als ein möglichens Modell unter allem andere. Diese Methoden werden als Hypothese-Tests nicht konstruiert, um die Alternativen zurückzuweisen. Wir werden diese zwei Methoden im Änderungspunkt-Entdeckungsproblem und anderen Ordnungsbeschränkungsproblem vergleichen. Auch werden wir ein modifiziertes Informationskriterium präsentieren, das die FWER-Kontrolle für die Musterauswahl unter der bestimmten Ordnungsbeschränkung kontrollieren kann.
\newpage



\section*{Abstract}
% Replace each subsection below with the appropriate contents.
% Feel free to add more (sub)sections if you want. Write clearly
% and concisely.


In the fields of biology, epidemiology and pharmacology, how to find
a suitable model to describe the dose-response relationship among
several groups is a common question. For example, questions, such as
"Are nickel-refinery worker more likely to get lung cancer than
others?", "How do ages of the parents effect the spontaneous

abortion rate?", "Is there any structure change between motifs and
'junk-DNA' on the DNA sequences?" and "Should we set standards for

the particulate matter in the air?If the answer is yes, then how
much?" are frequently asked.


Model-based \citep{Royston1999} and test-based \citep{Dosemeci1998}
approaches are most commonly used to analyze dose-response
relationship. In the questions mentioned above, the null hypothesis,
which usually assumes no difference or no change, should be tested
first. After the null hypothesis is rejected at a certain global level
$\alpha$, a model selection procedure is used to detect the
relationship. However, the test and model selection procedure
afterwards depend seriously on the dose-response shape, which is
unknown before the whole procedure. In this situation, a procedure,
which has good power over the alternative space, is required. An
order restricted likelihood ratio test is developed by
\cite{Robertson1988}. Other than the single test, a multiple
comparison procedure is developed by \cite{Mukerjee1987} who also
recommended orthogonal contrasts which have a simple power function.


The use of order-restricted hypotheses is a common approach which
increases power by narrowing the alternative space. Hereby,
Change-point, Simple-order and Simple-tree are three common types of
order restrictions. Several approaches are available for these
problems, such as max-t statistics according to \cite{Hirotsu2000},
which can be formulated as maximum contrast approach belonging to
the broader class of Multiple Contrast tests (MCT). The disadvantage
of MCT is that it makes model selection after the global null
hypotheses is rejected. The local finding rate of the true model of
the alternative is low. Recently, \cite{Anraku1999}, \cite{Zhao2002}
and \cite{Ninomiya2005} developed information-criterion-based
log-likelihood method for model selection approaches under certain
types of order restriction. These methods have better finding rate
of the true model, but have no FWER-control. They treat the null
model as one of the possible model among all others. These methods
are not constructed as hypothesis tests to reject the alternatives.
We will compare these two methods in Change-point detection problem
and other order restriction problems. Also we will present a modified
information criterion which can control the FWER-control for model
selection under different order restrictions.

[ 本帖最后由 orionsnow 于 2008-10-2 14:58 编辑 ]

salalaha

我不是你这个方向的,所以不知道给的comments是否valid:

1. 在我方向的文章里,通常abstract都比你的短(一般1-2paragraph),而且不会在abstract里cite别人的文章.

2. Other than the single test, a multiple comparison procedure is developed by \cite{Mukerjee1987} who also recommended orthogonal contrasts which have a simple power function.

by和who用的不对因为\cite{Mukerjee1987}指文章而不是作者.如果你想说作者,你可以在citation前加,e.g. Mukerjee et al.

3. such as max-t statistics according to \cite{Hirotsu2000},

Remove "according to", also use xxx~\cite{} 否则你的citation有时会跑到另一行.

4. \cite{Anraku1999}, \cite{Zhao2002}
and \cite{Ninomiya2005} developed information-criterion-based
log-likelihood method for model selection approaches under certain
types of order restriction.

一般不说\cite{} developed..., 要么说某某人developed,要么说某个approach被developed~\cite{...}.

5.你这个是survey paper吧,因为abstract基本没怎么谈你的东西,都是在比较别人的.

[ 本帖最后由 salalaha 于 2008-10-2 21:47 编辑 ]

foxtail_lb

我觉得德文的部分，小词太多了。不知道这个文章是发在什么杂志，或者网站上的？还有感觉德文的英文痕迹很厉害。hohohohohoho。如果发在德文杂志上，或者给很德国的人看，他们喜欢结构在复杂一点的句子。

俺是英文德文中文，哪个都做不到editor的地步，还是要请你的导师帮写一下。

orionsnow

$送花$ $送花$

xiexie,  

我的论文就是比较两大流派的方法，并且找到联系，补足缺点。

所以你看到前边就是很大部分在分析这两大流派。不过我会修改下，把上边这句填上

华丽

第一句德文译文完全翻译错了。句子结构和语法同英语的相比完全颠倒了，而且不通顺。结构应该是：Ein passendes Modell der Dosis-Wirkung Beziehung [....] zu finden ist eine allgemeine Frage in den Feldern der Biologie  ....

orionsnow

section*{摘要}
% Replace each subsection below with the appropriate contents.
% Feel free to add more (sub)sections if you want. Write clearly
% and concisely.


In the fields of biology, epidemiology and pharmacology, how to find
a suitable model to describe the dose-response relationship among
several groups is a common question. For example, questions, such as
"Are nickel-refinery worker more likely to get lung cancer than
others?", "How do ages of the parents effect the spontaneous
abortion rate?", "Is there any structure change between motifs and
'junk-DNA' on the DNA sequences?" and "Should we set standards for
the particulate matter in the air?If the answer is yes, then how
much?" are frequently asked.

在生物，生态学和医药领域，如何找到一个合适的模型来描述（药剂）
在不同的实验群体之间的量效关系，是一个很常遇到的问题。比如说，以下问题就经常被问及
：“镍矿提纯工人是不是更容易得肺癌？“，“父母的年龄是怎么影响流产率的？“，
“在DNA序列上，启动子和“垃圾DNA"之间有明显的结构变化么？“ 和“我们应该控制空气中胶粒体的含量么？“



Model-based \citep{Royston1999} and test-based \citep{Dosemeci1998}
approaches are most commonly used to analyze dose-response
relationship. In the questions mentioned above, the null hypothesis,
which usually assumes no difference or no change, should be tested
first. After the null hypothesis is rejected at a certain global level
$\alpha$, a model selection procedure is used to detect the
relationship. However, the test and model selection procedure
afterwards depend seriously on the dose-response shape, which is
unknown before the whole procedure. In this situation, a procedure,
which has good power over the alternative space, is required. An
order restricted likelihood ratio test is developed by
\cite{Robertson1988}. Other than the single test, a multiple
comparison procedure is developed by \cite{Mukerjee1987} who also
recommended orthogonal contrasts which have a simple power function.

基于模型识别的方法和基于假设检验的方法是处理量效关系时使用的两种常见方法。在我们上边提到的问题中，
（如果我们使用假设检验），零假设往往是，假设没有区别或没有变化。零假设需要先被检验，
在零假设以alpha 的概率被拒绝后，模型识别的方法再被应用到关系选择中来。但是这里有个问题，不管是用
模型识别的方法还是假设检验的方法，他们的严重的依赖于量效关系。而这个量效关系在没有作整个识别过程之前是不知道的。
所以我们需要一种方法，它需要在各种量效关系下都有很好的模型识别率。\cite{Robertson1988}建立了有序似然率估计法。
其他方法还有单项比对测试法， \cite{Mukerjee1987} 把这个方法发展成为了多项比对测试法，
他还建议我们用正交的比对项来获得比较简单的势函数。


The use of order-restricted hypotheses is a common approach which
increases power by narrowing the alternative space. Hereby,
Change-point, Simple-order and Simple-tree are three common types of
order restrictions. Several approaches are available for these
problems, such as max-t statistics according to \cite{Hirotsu2000},
which can be formulated as maximum contrast approach belonging to
the broader class of Multiple Contrast tests (MCT). The disadvantage
of MCT is that it makes model selection after the global null
hypotheses is rejected. The local finding rate of the true model of
the alternative is low. Recently, \cite{Anraku1999}, \cite{Zhao2002}
and \cite{Ninomiya2005} developed information-criterion-based
log-likelihood method for model selection approaches under certain
types of order restriction. These methods have better finding rate
of the true model, but have no FWER-control. They treat the null
model as one of the possible model among all others. These methods
are not constructed as hypothesis tests to reject the alternatives.
We will compare these two methods in Change-point detection problem
and other order restriction problems. Also we will present a modified
information criterion which can control the FWER-control for model
selection under different order restrictions.

使用有序假设有助于缩小否定域从而增加检验的势，在建设检验领域中这种做法被广泛应用。
在这里变化点，普通有序，普通树型有序，是三种常见的有序结构。有很多方法可以处理这些有序结构。
比如\cite{Hirotsu2000}的max-t 统计量，这种统计量可以被看成是多元比对测试的渐进最大值。
对于多元比对方法来说，它最大的弱点就是需要先拒绝零假设，然后才可以做模型识别。所以它的模型识别率不高。
最近\cite{Anraku1999}, \cite{Zhao2002}和 \cite{Ninomiya2005}
发展了一种新的基于信息论的对数似然法用来解决有序模型的识别问题。这些个方法有比较高的模型识别率。
但是它们不能控制犯总Ⅰ型错误（FWER）的概率。因为它们把零假设也当成了一个模型，和其他模型同等对待。
这些方法在被设计出来，目的不是在于控制错误概率的。我们在接下来的文章中比较这两种方法的优缺点。
并且给出它们在变化点有序结构和其他有序结构中的应用实例。我们还要展示一种新型的基于信息论的有序模型
识别法，这种方法既很好的选择了我们需要的模型，又有效的控制了总Ⅰ型错误的概率。