This the end day of 2003, tomorrow is a new year.
New year, will be new vision.
I'll try again!
2003年12月30日
First Volume of FSNLP
Last midnight, I finished the reading of the first volume of FSNLP.
This morning, I changed it for the second volume.
The first book gives me a summary of FSNLP. And I have found out that FSNLP is very good for the primary researcher to read for being familiar with NLP.
With the Adding Knowledge Program, I have persisted and gained a little.
I will keep on.
This morning, I changed it for the second volume.
The first book gives me a summary of FSNLP. And I have found out that FSNLP is very good for the primary researcher to read for being familiar with NLP.
With the Adding Knowledge Program, I have persisted and gained a little.
I will keep on.
2003年12月29日
YOCSEF Meeting
This morning, I arrived at the administration building's YOCSEF assembly room.
After we had disposed the assembly room, the meeting's chairman Pro. Zhao Tiejun declare the begin of YOCSEF. This is my second time to attend YOCSEF.
The subject of this YOCSEF is Digital Olympics and muitillanguage information processing. And there were three specially invited guests. They described the blueprint of Olympics and the current most difficulties.
I have been familiar with the general picture of the Digital Olympics. And I found out that the Information Processing is more useful in the current times. And the prospect of our lab is very beautiful.
After we had disposed the assembly room, the meeting's chairman Pro. Zhao Tiejun declare the begin of YOCSEF. This is my second time to attend YOCSEF.
The subject of this YOCSEF is Digital Olympics and muitillanguage information processing. And there were three specially invited guests. They described the blueprint of Olympics and the current most difficulties.
I have been familiar with the general picture of the Digital Olympics. And I found out that the Information Processing is more useful in the current times. And the prospect of our lab is very beautiful.
2003年12月28日
2003年12月27日
2003年12月26日
Rough Set Theory
Our WSD research group will do some experiments on Rough Set. And Mr.Lzm believes that rough set can be used for WSD very effectively.
So I am arranged to read dome materials on RS.
Firstly, I read Knowledge Discover by TSinghua Express. But the content of this book is not enough. So I find some papers about RS.
I have understood that reading summarize article is the fast way to be familiar with the area. I read a summarize paper about RS. And the knowledge points are very clearly.
So I am arranged to read dome materials on RS.
Firstly, I read Knowledge Discover by TSinghua Express. But the content of this book is not enough. So I find some papers about RS.
I have understood that reading summarize article is the fast way to be familiar with the area. I read a summarize paper about RS. And the knowledge points are very clearly.
2003年12月25日
Studying paper of Li Juanzi
This morning I am reading Li Juanzi's paper 《语言模型中的一种改进的最大熵方法及其应用》which is published on JOURNAL OF SOFTWARE.
In her paper, she used an updated method combining maximum entropy, mutual information and Z-test to choose the best feature of context for a multivocal word and then used IIS algorithm to optimize the parameters of the linear model for Word Sense Disambiguation.
The experiment results are displaying the advantage of this approach. But I think the paper has two flaws. Firstly, the experiments for WSD is not enough. Secondly, Z-test is usually used to test normal distribution for large scale's samples. And in this paper there is a connotative hypothesis that the mutual information between the feature set and the category a multivocal word is followed normal distribution. The experiments did not prove this hypothesis.
And I think this experiments could be done more fully, and the experiment should do the hypothesis test.
OK. There is an idea. We can arrange some little experiments of our current information corpus to do some hypothesis test. En, this idea should be thought more and more.
In her paper, she used an updated method combining maximum entropy, mutual information and Z-test to choose the best feature of context for a multivocal word and then used IIS algorithm to optimize the parameters of the linear model for Word Sense Disambiguation.
The experiment results are displaying the advantage of this approach. But I think the paper has two flaws. Firstly, the experiments for WSD is not enough. Secondly, Z-test is usually used to test normal distribution for large scale's samples. And in this paper there is a connotative hypothesis that the mutual information between the feature set and the category a multivocal word is followed normal distribution. The experiments did not prove this hypothesis.
And I think this experiments could be done more fully, and the experiment should do the hypothesis test.
OK. There is an idea. We can arrange some little experiments of our current information corpus to do some hypothesis test. En, this idea should be thought more and more.
2003年12月24日
2003年12月23日
A bless mail from Korean
One of my last year's MCM teammate, who is studying for her graduate degree in Korean Puxiang University, mailed a Christmas card to me.
She told me a lot of news of her study and life. She has gained the usually first achievement of her class. And at this Spring Festival she won't come back. Inseadly she will go to Seoul University to meet her friends. She has a long term scheme to publish two or more international papers. To this scheme she has a lot of confidence. And I believe her can do so.
Also, I have told my recent state and my works in recent months to her. She told me to study more studious and do more practices and prepare to do more achievement.
I make a good bless to her study and life in Korean and may her journey to Seoul.
She told me a lot of news of her study and life. She has gained the usually first achievement of her class. And at this Spring Festival she won't come back. Inseadly she will go to Seoul University to meet her friends. She has a long term scheme to publish two or more international papers. To this scheme she has a lot of confidence. And I believe her can do so.
Also, I have told my recent state and my works in recent months to her. She told me to study more studious and do more practices and prepare to do more achievement.
I make a good bless to her study and life in Korean and may her journey to Seoul.
2003年12月22日
A exciting news!!
There is a exciting news that our lab has gained first in a recent evaluating.
I'm excited. Dr.Tliu tells us that we should have self-confidence to do everything better.
Great!
I'm excited. Dr.Tliu tells us that we should have self-confidence to do everything better.
Great!
2003年12月21日
Matlab 程序升级
卢老师让我再做10-10，3-3，3-7，7-3等神经网络下的词义消歧实验。
想到上次5-5的实验花了我一整天的时间，而且还需要人工不断的切换，非常麻烦。
这次实验的总量是5-5的4倍，肯定不能采用原先的那种半自动的方法了。为此我花去了好几个小时来编制全自动实验的Matlab程序。
磨刀不误砍柴功。我在下午4：30编写完程序并将10-10的语料全部规范化后开始在服务器上运行程序，伪词01的21组实验花去三个小时左右就完成了。
这让我再次体会到了程序全自动化的好处。比起上次不断的人工切换程序，这次没有花费任何中途的人工干涉，效率大幅度提高亚。
想到上次5-5的实验花了我一整天的时间，而且还需要人工不断的切换，非常麻烦。
这次实验的总量是5-5的4倍，肯定不能采用原先的那种半自动的方法了。为此我花去了好几个小时来编制全自动实验的Matlab程序。
磨刀不误砍柴功。我在下午4：30编写完程序并将10-10的语料全部规范化后开始在服务器上运行程序，伪词01的21组实验花去三个小时左右就完成了。
这让我再次体会到了程序全自动化的好处。比起上次不断的人工切换程序，这次没有花费任何中途的人工干涉，效率大幅度提高亚。
2003年12月20日
2003年12月19日
Visual C++
In yesterday's adding knowledge program, I only read some base knowledge of Visual C++. But this evening, I did a small program on MFC after the guide of the book.
After I understood some codes of this program I was very happy.
MFC in .NET is more wonderful. Because it's desktop program like visual basic.
Great, study for program must do a lot of practices. I'll keep on.
After I understood some codes of this program I was very happy.
MFC in .NET is more wonderful. Because it's desktop program like visual basic.
Great, study for program must do a lot of practices. I'll keep on.
2003年12月18日
My Adding Knowledge Program
Up to now I have joined in IRLab for nearly five months. And I gradually discovered my shortage of my knowledge system. I summarized it into three main aspects: English general ability, Visual C++'s programming ability and the foundation knowledge of Natural Language Processing.
I had been analyzing them for a whole week. After the relative systemic analysis I made the Adding Knowledge Program aiming at the three aspects.
After compute everyday studying time, I was somewhat astonished. Because it is five hours. And after my careful thought, I found it was feasible.
The detailed program was made sure yesterday evening. And I carried it out this evening. Now I find it is reasonable for me.
I like the program. I believe I can stick to it.
I had been analyzing them for a whole week. After the relative systemic analysis I made the Adding Knowledge Program aiming at the three aspects.
After compute everyday studying time, I was somewhat astonished. Because it is five hours. And after my careful thought, I found it was feasible.
The detailed program was made sure yesterday evening. And I carried it out this evening. Now I find it is reasonable for me.
I like the program. I believe I can stick to it.
2003年12月17日
Deep research of our method of WSD
Our method has achieved to a good effect. But in our paper there are a lot of view points should be deep researched.
Dr.Tliu discussed detailed with Mr.Lzm and me. And we got a lot of constructive results which could lead us to do deep research of this point for WSD.
I should do some constructive experiments for our research.
Just so.
Dr.Tliu discussed detailed with Mr.Lzm and me. And we got a lot of constructive results which could lead us to do deep research of this point for WSD.
I should do some constructive experiments for our research.
Just so.
2003年12月16日
Dr. Rahmat Shoureshi's visit
This afternoon, about 4 o'clock, the American guest Dr. Rahmat Shoureshi visited our computer department. When he, with our dean, went into our lab, I made the demos for him.
I had being prepared for his visitation for three hours. But I had somewhat strain. When I made the demo of Dependency Parsing, the visitor maybe be confused. So Dean Xu let me change another one. I made the latter demo of Englishing Writing Assistant. I said a lot of prepared sentences to him. At last, he understood the main idea of the demos.
I am very happy to this chance for communicating with this Doctor.
I had being prepared for his visitation for three hours. But I had somewhat strain. When I made the demo of Dependency Parsing, the visitor maybe be confused. So Dean Xu let me change another one. I made the latter demo of Englishing Writing Assistant. I said a lot of prepared sentences to him. At last, he understood the main idea of the demos.
I am very happy to this chance for communicating with this Doctor.
2003年12月15日
Rough Set theory
Today I read through the paper of Dr.Chen Qingcai. In his paper rough set theory has been used very often.
It can be used in changeing pinyin into words, and reducting the baseline words for computing similitude degree of words.
A doctorial paper is very ample. I have found the difficult of it.
But at the same time I find out that grey system, like rough set, can be used widely into Chinese Information Processing. But the foundation of Chinese Information Processing and Grey system should be combined. In order to do so, I should learn more of them.
Try, and again.
It can be used in changeing pinyin into words, and reducting the baseline words for computing similitude degree of words.
A doctorial paper is very ample. I have found the difficult of it.
But at the same time I find out that grey system, like rough set, can be used widely into Chinese Information Processing. But the foundation of Chinese Information Processing and Grey system should be combined. In order to do so, I should learn more of them.
Try, and again.
2003年12月14日
2003年12月13日
Get along with the experiment
At 8:30 I began to do the one hundred and sixty-eight experiments. I divided the experiments into four parts. I followed the way which I had used in the science and technology innovation contest that change the iterate times to observe the trend of results.
I implemented them one by one, and wrote the results on my notebook.
At 4:30 this afternoon, I had done them finally. At that time I was very happy.
Mr.Lzm observed the results, and asked me to study the paper of Dr.Chen Qingcai and try to use the rough set theory.
Rough set is very good theory. I think it is useful as Grey system theory. I will contrast them and use them in WSD.
I implemented them one by one, and wrote the results on my notebook.
At 4:30 this afternoon, I had done them finally. At that time I was very happy.
Mr.Lzm observed the results, and asked me to study the paper of Dr.Chen Qingcai and try to use the rough set theory.
Rough set is very good theory. I think it is useful as Grey system theory. I will contrast them and use them in WSD.
2003年12月12日
Detailed experiment for NN-WSD
Based on the result of yesterday's full xperiments, after discussing the detaild experiment with Mr.Lzm, I begin to do the series of experiments on all kinds of corpor scales.
The diffilcult portion of the series experiments is to chage the original corpus into the format for matlab.
After I had done the difficult portion, I implement the programms and get the results quickly.
When I hand over today's results to Mr.Lzm, he gives me a lot of suggests for updating the experiment. So I will do careful experiments on five cmputers tomorrow.
It will be taken a whole day.
The diffilcult portion of the series experiments is to chage the original corpus into the format for matlab.
After I had done the difficult portion, I implement the programms and get the results quickly.
When I hand over today's results to Mr.Lzm, he gives me a lot of suggests for updating the experiment. So I will do careful experiments on five cmputers tomorrow.
It will be taken a whole day.
2003年12月11日
Full Experiment of WSD
In order to make sure the optimal inside net structure, I have done a self-contained experiment plan.
Likely to the experiment of my science and technology innovation, I confirm the parameters one by one.
At last, just now, I complete the whole experiment plan. I make the conclusion that: the magnify parameter is 9, the err_goal is 0.3, one connotative layer is enough and obligatory, the number of the connotative layer's node is 12.
I thought originally that two connotative layers is the best choice. But after my test experiment, I find out my former thought is wrong.
Experiment is the best tool to prove your idea.
Likely to the experiment of my science and technology innovation, I confirm the parameters one by one.
At last, just now, I complete the whole experiment plan. I make the conclusion that: the magnify parameter is 9, the err_goal is 0.3, one connotative layer is enough and obligatory, the number of the connotative layer's node is 12.
I thought originally that two connotative layers is the best choice. But after my test experiment, I find out my former thought is wrong.
Experiment is the best tool to prove your idea.
2003年12月10日
O-O for Software Engineer
This evening, I begin to study O-O for Software Engineer.
At the class, I find out that the teacher's style is very different from Chinese teacher. Because the teacher, with five years oversea working career, is back from America. After the discuss of my friend and me, we agree with each other. We think that American teaching style is to make the problem simplier and simplier. The questions we are asked are very easy and can be answered quickly. Their teaching contains a lot of instances and is short of clearly logic.
At the end of the class, I begin to adapt to her teaching style.
Great！ Continue studying for it and going to the classes.
At the class, I find out that the teacher's style is very different from Chinese teacher. Because the teacher, with five years oversea working career, is back from America. After the discuss of my friend and me, we agree with each other. We think that American teaching style is to make the problem simplier and simplier. The questions we are asked are very easy and can be answered quickly. Their teaching contains a lot of instances and is short of clearly logic.
At the end of the class, I begin to adapt to her teaching style.
Great！ Continue studying for it and going to the classes.
2003年12月9日
2003年12月8日
《自然语言理解-计算机能思维吗》
昨天开始学习王开铸老师在1995年写的《自然语言理解-计算机能思维吗》，书很薄，但很精辟。我感悟颇深。
摘抄一些如下：
经过漫长的社会演变，已经形成如今的八大语系：汉藏语系、印欧语系、亚非语系、阿尔泰语系、乌拉尔语系、尼日尔-刚果语系、马来-玻里尼西语系和德拉维达语系。
自然语言理解的三种观点：系统工程观点、层次结构观点、层次间单向观点。
对话双方的言语链过程：思维层-〉生理层-〉物理层-〉生理层-〉语言层-〉思维层
以上这些均是从第一章中摘录的。 仔细理解，确实很耐人寻味。
摘抄一些如下：
经过漫长的社会演变，已经形成如今的八大语系：汉藏语系、印欧语系、亚非语系、阿尔泰语系、乌拉尔语系、尼日尔-刚果语系、马来-玻里尼西语系和德拉维达语系。
自然语言理解的三种观点：系统工程观点、层次结构观点、层次间单向观点。
对话双方的言语链过程：思维层-〉生理层-〉物理层-〉生理层-〉语言层-〉思维层
以上这些均是从第一章中摘录的。 仔细理解，确实很耐人寻味。
2003年12月7日
The adjusting period
This afternoon our lab has the weekly report meeting. It is turn to our Dic-Constructing group to report our works and difficult.
Zhu Liuliu take the report firstly. And I give two demos of our Cup-Dic. Indeed, there are a lot of shortages of our data files and demos. And I have the confidence to make it better, under the guide from Dr.Tliu and Mr.Lzm.
If you find the errors, you can solve them one by one. Yes, our group has found the shortages of our work. We will cut them one by one. So, I think we will do better after some time.
And after these days busy, I find some lack of my work style and arranging of my life. I analysis and get the conclusions as follows.
Firstly, my attention is easy to be thrown into confusion. If there is a emergent task which I must finish in few days, I easily do not carry out my former plan.
Secondly, my emotion is impacted easily by the current working state. Before a week, I should finish three big tasks parallel in few days. During my hard and hard working, I was not happy. But after the busy period I find I can do them better if I plan better.
Sharpen the saw. I must plan carefully enough, and do them one by one every day. Every morning, I should plan my detailed tasks of this day. At evening, I must chek them one by one, and adjust the leaves. Every week, I should do like this. So do every month.
Do them at once.
Zhu Liuliu take the report firstly. And I give two demos of our Cup-Dic. Indeed, there are a lot of shortages of our data files and demos. And I have the confidence to make it better, under the guide from Dr.Tliu and Mr.Lzm.
If you find the errors, you can solve them one by one. Yes, our group has found the shortages of our work. We will cut them one by one. So, I think we will do better after some time.
And after these days busy, I find some lack of my work style and arranging of my life. I analysis and get the conclusions as follows.
Firstly, my attention is easy to be thrown into confusion. If there is a emergent task which I must finish in few days, I easily do not carry out my former plan.
Secondly, my emotion is impacted easily by the current working state. Before a week, I should finish three big tasks parallel in few days. During my hard and hard working, I was not happy. But after the busy period I find I can do them better if I plan better.
Sharpen the saw. I must plan carefully enough, and do them one by one every day. Every morning, I should plan my detailed tasks of this day. At evening, I must chek them one by one, and adjust the leaves. Every week, I should do like this. So do every month.
Do them at once.
2003年12月6日
Construction of the Dictionary
This weekend, it is turn to Zhu Liuliu and me to report the latest development of the construction os the dictionary.
Zhu Liuliu and I were preparing the materials for the report. I have outlined the report into two mainparts. First one, for Zhu Liuliu, are bottom data files construction and the similitude degree of Chinese words, and the other one, for me, are the two demos of IdeaNet, Cup-Dicthe shortages and the next scheme.
I have done my part. And just Zhu Liuliu have done,too.
Zhu Liuliu and I were preparing the materials for the report. I have outlined the report into two mainparts. First one, for Zhu Liuliu, are bottom data files construction and the similitude degree of Chinese words, and the other one, for me, are the two demos of IdeaNet, Cup-Dicthe shortages and the next scheme.
I have done my part. And just Zhu Liuliu have done,too.
2003年12月4日
Exciting of the message from Dr.Tliu
These days our lab is devoting for the Big project of National center.
And Dr.Tliu gave us some very exiting news.
This afternoon, Dr.Tliu came into our room with the successing message.
All of us were excited.
Yes, it is exciting. We should try more！
And Dr.Tliu gave us some very exiting news.
This afternoon, Dr.Tliu came into our room with the successing message.
All of us were excited.
Yes, it is exciting. We should try more！
2003年12月3日
2003年12月2日
Exam of S.E.
Software Engineering is the last exam of the classes that should be numbered in the total achievement.
I have been reviewing for it for five days. I read through the book at least two times. And I have done one exam paper and the all homework to Mrs.Wang Yuying's powerpoint.
I think I have prepared fully for it.
But when I doing my exam paper, I find there is a lot of difference. The paper include some knowledge about the Compile Theory and the whole system of our school. I add a lot of my experience of my development of some software.
I think I have tried my best. It is enough.
I have been reviewing for it for five days. I read through the book at least two times. And I have done one exam paper and the all homework to Mrs.Wang Yuying's powerpoint.
I think I have prepared fully for it.
But when I doing my exam paper, I find there is a lot of difference. The paper include some knowledge about the Compile Theory and the whole system of our school. I add a lot of my experience of my development of some software.
I think I have tried my best. It is enough.
2003年12月1日
The last class of Communist Party Colledge
This evening, there is the last class of Communist Party Colledge.
After the compere's audios of Divine Boat and Yang Liwei, some students go to the dais to talk about the self-feeling.
I am the third one who go to the dais. Firstly, I talk about the Diving boat success's significance. Then I state my viewpoint about the Diving boat's success. I anlysis the contrast between little success of Diving boat and the austere situation of our nation. I think the success of Diving boat is very little of the full project of our nation's renaissance, and we must foucs on the other things of the face of the project.
But the career is not by only day, it is a long term scheme. We must try our best to do what we must do and what we want to do.
After my address, all of our students agree with my view.
It is ok!
After the compere's audios of Divine Boat and Yang Liwei, some students go to the dais to talk about the self-feeling.
I am the third one who go to the dais. Firstly, I talk about the Diving boat success's significance. Then I state my viewpoint about the Diving boat's success. I anlysis the contrast between little success of Diving boat and the austere situation of our nation. I think the success of Diving boat is very little of the full project of our nation's renaissance, and we must foucs on the other things of the face of the project.
But the career is not by only day, it is a long term scheme. We must try our best to do what we must do and what we want to do.
After my address, all of our students agree with my view.
It is ok!
订阅：
博文 (Atom)