這篇文章主要是改打分系統的評測。託福雅思高考等https://app.mtutor.engkoo.com/englishwriting/，gre見最下面。
另外人工智能的意思是，微軟小英會根據你的文章內容幫你改作文，比如他能補全你缺少的相關名詞。
目錄
如果chrome顯示打分失敗換別的瀏覽器比如edge，我就出現了這個問題
先展示下評分功能：它可以從四個方面給出接近託福的打分。
比如，題目是：
Do you agree or disagree with the following statement? All university students should be required to take history courses no matter what their field of study is. Use specific reasons and examples to support your answer.
多種考試類型可以選擇：
如果修改後再要求打分：
主要特點：
1.打分很快
2.能夠識別拼寫錯誤
3.能夠識別語法錯誤，包括動名詞等，以及一些比較高級的語法錯誤
4.能夠自動修正
5.感覺對於高級詞意理解可能不是那麼合適，他不會告訴你哪個單詞用的不恰當，所以建議配合谷歌翻譯使用。
如果chrome顯示打分失敗換別的瀏覽器比如edge，我就出現了這個問題
人工外教批改
然後吧，這個30分作文我用了同桌英語的vip批改，給了我3分檔（0-5），相當於託福的20分（0-30）.
然後評語是：
3分對應的：
♦可能因完全曲解寫作任務而未能根據寫作任務行文
♦所表達的觀點有限，可能大部分並不相關或重複
♦不能有邏輯地組織觀點
♦所用銜接手段有時非常有限，且有時未能體現觀點之間的邏輯性
♦ 只使用非常有限的詞彙及表達方式，對構詞及/或拼寫掌握也非常有限
♦ 錯誤可能嚴重地影響信息的傳達
♦ 嘗試造句，但語法及標點符號錯誤佔多數，意思被扭曲
感覺是說少了結尾段，我個人不喜歡寫結尾，覺得廢話，開頭我個人也不喜歡重複題目，覺得廢話。哈哈。寫作習慣看來不適合考試。每個段的結尾也要重複一邊結論，就感覺很不實際，湊字數一樣。模板意味很重。看來考試有考試的套路。
另外就是論點，他認爲不夠充分，我個人也不喜歡舉例我堂弟，我同學，我朋友等。比較喜歡推邏輯。
不能偏聽偏信，不過也要吸取別人的建議。寫一篇模板比較保險。
30分/3分原文：
I disagree with the opinion that all college students should take history classes for a few reasons.
First, college students major subject is already stressful, especially with goals of a straight A. They have to go to classes on time and finish all of their homework assignments. To impress the professors they also need to preview a lot of materials relevant or not to the classes. For example, my major is computer science and we have code missions, but sometimes the bug will take days to be corrected. If there are irrelevant subjects such as history to give us additional burdens, our college life will be really tough.
In addition to the stressful study, enthusiasm is another reason that students in other fields should not learn history. They did not chose history as their major at first , it might be that they are not interested in history, and some might hate it. Forcing students in history , as history needs passions to remember, might lead to a backfire situation easily, hence the more school wants them learn the less they will remember. Consequently, they might fail and have to learn it again.
So is it harmful to your life knowing nothing about history? Absolutely not. Because different countries have different histories. It is normal to know nothing about the history of a country. For example, nowadays people move and immigration a lot, and for them general knowledge of a country is enough. So why bother to learn history, if you do not need it.
Last but not least, we can consider this learning problem from teachers' and school's perspectives as well. All the students taking history classes will certainly requires more teachers. So the school would rather have teachers in class longer than before or hire more history teachers. The first option will consume too much energy from teachers, which is responsible for the low effects in classes and limited research from professors. And the second one will spend teaching funds on a program which nobody likes, generate zero benefit and undermine research.
同桌英語改後：放圖片主要是因爲能夠看的很清楚加了主要的是一些總結句。實際上內容是不給你改的，大概還是需要自己琢磨怎麼豐富細節。不過在通過人工智能修改後，大部分的語法錯誤，都沒了。
有一段外教以爲我用的使轉折，實際上使遞進，就很尷尬，看來寫的應該更加直白一些。
GRE
gre有官方的打分項目，不過兩篇文章20大洋，美刀。相當於一篇80人民幣。怕了怕了。
所以這裏放一個在線人工智能打分的項目。https://www.mbacrystalball.com/gre/gre-essay-grader
也是自然語言處理處理的。
可以先通過小英修改，然後再用評分系統進行打分。
我測試了一篇滿分範文，得到了5.7的分數，看來這個系統其實是比較保守的。所以我得了5.8是不是說明溢出了。開個玩笑，初期寫作花了1小時認真寫，又花了兩個小時各種改，纔得到這個分數，還是要認真的練習。
比如下面這篇文章，我就是同過各種人工智能方式改的語法錯誤，最後再用人工智能評分，喫土，
The data generate an argument that the less homework makes Marlee students achieve more grades in math and science than Sanlee where students get more homework. But the conclusion is based on three wrong premises : First, the students may be too lazy to finish the works which are accordingly not responsible for the bad scores. Second, the school can set different test and graduation standards so the grades are not actually comparability. Third, the frequency does not equal to the quality and quantity with regard to homework thus students in Marlee may need to spend more time and energy on their difficult task.
To begin with, the argument based on a unwarranted premise that the students will review all the knowledge from homeworks industriously, which contains finishing all of them timely, correcting them with the answers as soon as they finish and reviewing the wrong one frequently. But if the students just ignore most of their homework, it will be default. For example, my father is a clever man, but he did not make it to college. Also, he always reminds me to treat study seriously. Turns out he used his intelligence to escape every work and class in high school. As the consequence showed by my father, if the homework are not done, laziness might be the chief culprit of low scores.
Additionally, the argument supposes that schools in two districts use the same test and graduation standards, so the same scores mean students in Marlee are outdo those in Marlee. But it is normal that good schools set difficult quizzes so that their students can compete with each other and detect the weakness through the incorrect ones. For them, all students get full marks when test in other schools. For example, Tsinghua is the top university in China, and I took their entrance exam before. I was not surprised that everything in it reached the highest difficulty that I have ever seen. On the contrary , after I had failed the test in Tsinghua and taken another one in other schools, I got high marks easily. So do the different effects of various graduation standard. In one word, the score does not reflect the ability of students unless they take the same exam and those in Sanlee may have higher scores if they take the tests of Marlee.
Last but not least, the argument seems to take the frequency of homework as the workload. A few days a week is truly less than everyday, but if the task for a few days is complicate and consist of several small parts then the students in Marlee actually finish more workloads on study to get good score. For example, college students usually get homework once a week, but it doesn't mean that they have a easy work, because their essay task contains lots of time of searching material, thinking and discussion, and entity homework like computer code will take time to code and debug. As for the quantity, some teachers will even set homework for a month so the students can preview and finish them before classes, then they can understand better. So the homework with low frequency but high difficulty and quantity cast the positive affection on scores in Marlee.
As I mentioned above, the three premise is untenable thus the logic that the less homework in Marlee get higher marks than those in Sanlee is questionable.
GRE方法
我通過小英進行改錯，通過谷歌翻譯將文章英譯中，然後更改奇怪的語義不通的地方。最後使用下面的進行判分。
總結
就計算機科學角度來講，如果官方已經實現了機器打分，那麼非官方的機器打分有80%的可信賴程度，因爲代碼以及方法大半都是公開的，計算機比較講究開源。唯一可能導致打分不那麼準確的原因使數據庫的差異，但是這個差異根據一般論文來看，準確率會有最多5%的變動，一般都是2%。
人工批改顯然有更多的細節在內容上，因爲人工智能並沒有審題，也不會考慮內容的相識，直白，人性化。也沒有要求你的模板。所以可以在初期使用人工智能以及寫作模板達到一個較好的分數，如果想要高分滿分，顯然是需要購買一定的人工批改的。這並非恰飯，我還沒這本事，sad。