open mind

来源: BlogBus 原始链接: http://jlaky.blogbus.com:80/s8330/ 存档链接: https://web.archive.org/web/20061103123334id_/http://jlaky.blogbus.com:80/s8330/

open mind make a difference 首页 最后更新 最新评论 存档 我的链接 偏执 乱想 -[ thinking ] 压力越大越偏执，偏执则带来更大的压力。 我偏执是过分自由的兴趣导向引起的，一直以来，程序，引擎，工程，梦想与虚幻夹杂在一起。钱？我很在乎吗？只要够生活就行了。我最想要做的是什么，是一个真正意义的引擎。 幸福的轮廓往往隐藏的是毁灭，相反毁灭过后也能看到幸福的轮廓。上次在刚看到“这个幸福的轮廓就是偶然”，偶然某种意义上也许就是必然。而我生活在价值观极其单一的社会中，我也会进入那一族？抽着烟，端着美酒，抱着美女，握着股票，谈论钞票,还粗鲁的笑？无止境的金钱和欲望？ 幸福不是这样的吧。。。 相信，便得救。。。 jlaky 发表于 2006-09-21 00:19:49 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 DP与中文分词 -[ thinking ] 早上醒来就在想DP的使用范围似乎很熟悉，串匹配？中文分词？DP似乎可以解决中文分词的最佳分解问题？ google搜一下，wk，google黑板报就提到了，以前看这篇文章没有注意： 我们假定一个句子S可以有几种分词方法，为了简单起见我们假定有以下三种： A1, A2, A3, ..., Ak, B1, B2, B3, ..., Bm C1, C2, C3, ..., Cn 其中，A1, A2, B1, B2, C1, C2 等等都是汉语的词。那么最好的一种分词方法应该保证分完词后这个句子出现的概率最大。也就是说如果 A1,A2,..., Ak 是最好的分法，那么 （P 表示概率）： P (A1, A2, A3, ..., Ak） 〉 P (B1, B2, B3, ..., Bm), 并且 P (A1, A2, A3, ..., Ak） 〉 P(C1, C2, C3, ..., Cn) 因此，只要我们利用上回提到的统计语言模型计算出每种分词后句子出现的概率，并找出其中概率最大的，我们就能够找到最好的分词方法。 当然，这里面有一个实现的技巧。如果我们穷举所有可能的分词方法并计算出每种可能性下句子的概率，那么计算量是相当大的。因此，我们可以把它看成是一个动态规划（Dynamic Programming) 的问题，并利用 “维特比”（Viterbi） 算法快速地找到最佳分词。 jlaky 发表于 2006-08-16 09:33:23 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 幸福 -[ thinking ] 在异乡和老朋友见面聊天，是一种幸福。 每解决一个技术难题，程序跑出意料中的结果，是一种幸福。 看到好书的作者的观点和自己的观点相近或者一致的，或者被作者的观点点醒的时候，也是一种幸福。 在Topcoder相互切磋，以“程序”会友，是一种幸福。 在不经意间看到一个哲理或者规则，也是一种幸福。 每次经历痛苦煎熬的时候突然看透根源的时候，也是一种幸福。 睡觉睡到自然醒，也是一种幸福。 听到好听的歌，也是一种幸福。 幸福，如此简单，所以生活，顺其自然。 jlaky 发表于 2006-08-14 21:08:20 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 过犹不及 -[ thinking ] 信息量过大和信息不足都是问题，所以我们需要搜索，人的精力有限，人的时间有限，所以不可能吸取所有的知识，下载了大量的电子书有多少时间去看？或许我们真正需要的只是那么几句提点，书越来越厚，往往我们看不到精华就扔一边了。好书应该简短精确，无废话，设计模式就是很好的例子，如果更好就是容易理解并且简短，这两点往往是矛盾的。无论如何事无巨细就是往往是过犹不及的表现，浪费的是读者也浪费自己的时间。将来是否有一种工具能根据自身的特点选取特定的地方读取？也就是信息的个性化。个性化是减少信息冗余的有效方法。 jlaky 发表于 2006-08-14 12:02:56 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 第三代搜索？ -[ thinking ] http://news.csdn.net/n/20060807/93362.html 一旦发现有利可图，一堆人立马炒作概念 语义搜索？ 难道要每次搜索前先把句子成分标识清楚？ 那用户估计会烦死？ 而且我们一般很少会真找几万张网页， 而且翻几页找不到， 一般就会选其他关键字， 专业搜索引擎是一种缩小范围的解决问题的方法，很早就出现了。 不过目前我们大多数所需要的是通用的，而不是专业的，简单有效才是关键。 jlaky 发表于 2006-08-12 23:37:00 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 "三维"搜索引擎算法 准确度提高31%? -[ thinking ] 微软推"三维"搜索引擎算法 准确度提高31% from csdn 日前微软表示，旗下科学家已经开始着手研究下一代搜索技术，从而改变如今的搜索排名原理。 据news.com报道，一份微 软研究员提交的报告显示，新搜索技术将对“排名算法”进行改进，“如今绝大多数搜索引擎都在使用二位算法，将用户的搜索查询和后台网页连接直接挂钩，并且 返回搜索结果。而微软将创建一种全新的‘三维搜索’概念，为用户带来前所未有的搜索经历。”微软实验室数据挖掘、搜索、导航专家欧西先生如是说。 据悉，三维搜索引擎将分析“用户在某个网页的逗留时间，”除了点击率之外，再以逗留时间作为某个网页的排序标准。“通过全新的算法，我们将更为精确的了解用户都某个网站的评价和看法，比起之前的搜索引擎，新技术将提供更准确的搜索结果。” 在接下来一次试验中，研究人员对3000次网页搜索和1200万名搜索者进行了试验，结果表明新型搜索引擎算法在准确性上足足比旧方法提高了31％。另 一方面，在日前召开的第29界年度国际电脑机器特殊兴趣小组协作会议（ACM SIGIR 2006）上，来自于美国密歇根州大学的研究人员还展示了如何将“数字内容压缩”，从而更好的应对移动搜索平台。 //按照这种排序算法，除了PageRank另外加上网页的停留时间作为影响因子 //用户时间总和越长网站越重要？ //如何加权？ //如何评价试验？ //如何获得用户在某网页的停留时间？ //如何防止作弊？ //计算过程是否还足够快？ //感觉将问题复杂化了。 jlaky 发表于 2006-08-12 23:25:37 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 Algorithm is mind and technology is gun -[ thinking ] We can shoot down anything with them! jlaky 发表于 2006-04-22 13:36:09 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 Lesson -[ thinking ] Totally misled by two papers, I implemented the algorithms in my they describe and concluded that their don't try the algorithms in real system.But I waste so much time on the math equations.It is a lesson, don't believe the paper said, at least the more authoritive would be better. And then I think about the authority, it is good as being led by the authrity at first, as authority may be a collection of wisdom that stands for the research area and the direction.And if you are in the area for a long time and in depth, is it so important of the authority of a paper , maybe it will be more important that of good idea and do experiment to prove it. I noticed that many Ph.D here read a large number of papers, they read paper day by day, but seldom have their own mind.They seems like dogs that led by the bell sound, the bell rings, then it is time to find new approach, it is time to criticism their paper!And to post my paper in a more mature equation or hypothesis,It is really worthy wasting so much time?I think the most important thing of research is the prove and experiment rather than the mathematical criticism.The mathematical criticism will led to a circle to create more and more complex equations.Math is very important, and beautiful, but it is not means that the obscurer the better, it is superficial.I find it the paper or application of funding,if there is no obscure equation,it will be of less chance to publish or cited, less chance to get the funding. How to change the condition?It may be very dificult,it is hard to find a more reasonable standard. jlaky 发表于 2006-04-15 11:52:45 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 What a crazy city! -[ thinking ] We went to the west-tower district of the shit city today.And sth weird happended.The district seems Korean than Chinese, the stores and the restaurant are all named in Korean.And the price of the merchandise here is all much more expensive. We had the chicken soap in a restaurant, and each bowl will cost 30 yuan, more expensive than any where in China, at least more expensive than the central part of Beijing.And I heard that 13 people paid 3600 yuan for 10 dumplings each in the each street here.What a crazy city! And then we went to a supermarket for some drink, when we went to one check-out desk to paid for the drink, as there are much fewer queuer than other desk, but the sales girl show us that this desk is closed, but we just saw there were just 4 guys check out from here, what is the matter? And we went to another desk in a long queue, when there are four guys went to that desk, and the girl is happy to serve them to check out. Then we went back to there, and find that, the language they spoke is not Chinese but Korean.But the sales girl must be Chinese, when they check out, I noticed that they paid RMB not Korean dollar, so why we had to be treat differently? Why we have to take the long queue? The only reason is that we are Chinese! But this is in China! This fking place is so crazy!We took the national discrimination in China, our own country! But what can we do? Something unpleasant in my brain.Despairing of the research institutes in China, and now is the civilization now is corrupt, I am thinking what is matter in China? Even 2 months ago, I still believed the technology in China some day will exceed other opponents in Asia even in the world.But now, I find it will be very very hard if the honor is corrupt.The only thing for me to do is just do my best, and be smarter and smarter, stronger and stronger, I can find something different someday. jlaky 发表于 2006-04-09 20:39:54 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 thinkng about the google translater -[ thinking ] I find from the google blackboard that the basic arithmatic approach for google translater system is very simple. P(S) = P(w1)P(w2|w1)P(w3|w2)... And it is efficient in both research and practical problem, it works more efficiently than the rule-based system.Math is magic! But it should help us solve a problem in a simple but efficient way but not lead us to a complicated area.Every problem can be solve in a simple way, who can solve it is that who find it and do it. Image that a band of guys are there struggling to solve a problem using a complicated approach, and at last, they find it is so hard and give it up.The fact is that the problem is not difficult at all but the approach they use.I remember when I was sophomore in college, after the class of data structure, I was discussing a project problem with the professor, I argue that I can use a complex data structure to solve a problem but he just asked me a problem, you can solve it in a pretty simple way, why don you want to solve it in a much more complex way.I still insisted to my thinking. And than, when I implement the approach, I find it is so complex a problem, it take me several days to debug.But the classmates around solve it in a much shorter time.After that, I always think about a problem like : is there any approach can solve it easier, why I have to do it like this.But still, I am trapped by some silly approaches. Keep it easy, but strong enough to solve the problem! jlaky 发表于 2006-04-06 19:27:49 阅读全文 | 评论(2) | 引用Trackback(0) | 编辑 thinking about the work -[ thinking ] When I was very young, my parents told me, you should study hard so as to get a easy job with high standard of living, you should study because it will bring you a good work.So I study hard as they wanted. And remember the teacher in my high school, I knew that teachers seldom said they loved their job, and they just told us to study, you can go to good college for further study.When the first day I came to the college the teacher told me in my department, the graduates were very easy to find a work with high sallary, it would be very easy to make a living. If you get a good grade in the class, you would have more choices than just to work. And my classmates always talk about the sallary and the offer, I was all confused in a question that did I have to do this just because I want to make better a living?I was once very interested in how to make money by investment and how to find oppurtunity in stock market the Wall Street.When I take the part time job, I always met the terrible problem with procastination, in fact, I never do what I really want to do.When there is a time I develop the 51life.cn website, I always cursed the site, that is a reason I once gave up programming, as it is easy. And finally, I find that the original desire for the money is not what I want most but what I want to make by creating something.All of these implied a truth that to work is conflicted with to have fun, work and fun are all opposites by definition. Most of us have been thoroughly misled about the idea of loving one's work.School has trained them to regard work as an unpleasant duty. Having a job is said to be even more onerous than schoolwork. And yet all the adults claim to like what they do.But how to judge your work as the work you really want to do? Try to do something that make your friend say :wow, that's pretty cool. And you take pride of it. You shouldn't worry about prestige, as it is the opinon of the rest the world. "When you can ask the opinions of people whose judgement you respect, what does it add to consider the opinions of people you don't even know? " " So just do what you like, and let prestige take care of itself." "The other big force leading people astray is money. Money by itself is not that dangerous. When something pays well but is regarded with contempt, like telemarketing, or prostitution, or personal injury litigation, ambitious people aren't tempted by it. That kind of work ends up being done by people who are "just trying to make a living." " jlaky 发表于 2006-04-03 19:54:39 阅读全文 | 评论(1) | 引用Trackback(0) | 编辑 37signals: Programming At Warp Speed -[ thinking ] from http://businessweek.com/magazine/content/06_13/b3977005.htm : The lesson: Create a simple product as fast as you can, then get feedback from customers and make it better In fact, 37signals turns the old ways upside down. For starters, its applications are delivered on the Web for a monthly fee, not sold in boxes. The company's seven employees don't believe in planning, either. They just start creating and trying things out.They just start creating and trying things out. And rather than loading products with bells and whistles, they design them to do a few things well."The way to get really good software is to make the simplest thing you can as fast as you can and get reaction, then see where it goes from there," says Paul Graham, a pioneer in Web-based software and now a guru for software entrepreneurs who operate like those at 37signals There's a downside to this way of doing business. If one upstart can deliver a new product in a snap, so can others. The guys at 37signals aren't sweating it. In fact, between projects they travel the country giving seminars and promoting their book, Getting Real, about how they build applications. Unlike many dot-comers who followed the get-big-fast credo, Fried has no plans to add to his staff. The goal: Stay small and agile -- and keep pumping out those programs. A typical agile programming practice example.We can always make things better and easier.Simple version ->next version->next version , iterate in a short period of time,just keep it easy and agile, your software will be good.When I took my part time job in a small company, the architecture of their service platform is so complicated and full of bugs just like a nightmare. The last month, I was trying to persuade them to use some agile programming approach, but of no sense, they don't know it at all.That is a really "hardwork" place.Then I left. jlaky 发表于 2006-04-02 20:07:06 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 协调和对抗 -[ thinking ] 协调和对抗策略的问题其实是同步的。 其实就是合作问题，首先采用什么协调结构，协调首先需要一个info pool，包含所有我方信息和对方信息，而这些信息不能直接使用或者不需要所有都用，可以增加一层antagonism knowledge，就是从info pool进行深一步的挖掘得到知识。比如，对方如果控制球，正在向我方进攻，那么我方该如何防守？可以获取的信息包括我方和对方的位置和速度，带球队员的位置和方向。根据这个信息，获取对方下一步移到的位置，服务器的主要任务就是计算在什么位置可以截住进攻球员，这个是一种对抗方式，另外一种是传球和射门的拦截，进一步获取到传球的路线，估计拦截的位置，其他两个球员的跑位可以计算出来，其实难点就是如何计算跑位路线保证队友不相碰而且可以最好的完成任务呢？这个就是协调问题。协调的本质就是为了更好的对抗。用一种简单的概率方法来猜测对方的行动，然后策略针对对方最可能的举动来执行。 jlaky 发表于 2006-03-05 19:52:24 阅读全文 | 评论(1) | 引用Trackback(0) | 编辑 杂感 -[ thinking ] 发现为钱忙碌真的很容易疲惫，而钱又是生活的动力，所以很多人都很累，为理想忙碌更高一层，记得在实验室忙碌的一段时间，没有什么金钱诱惑，我为什么愿意花大量的时间进去？兴趣，就是兴趣。 但是兴趣不等于理想。 在中国，所谓成功，往往都是和钱地位相关的，也就是一元化模式，这个似乎是从小灌输的定势思维。 理想和成功往往是相关的。所有人都在为自己的某种目标而活着，达到目标可以看成是成功。大部分人的生活目的都是一样的富足而轻松的生活。那么理想，理想又是什么？得到某种程度的物质需求？ 宏观上物质基础决定社会形态，微观上经济也决定个人的意识。 这些都不足以支撑理想的重量。如果理想是无偿愿意花费大量时间为之奋斗的目标的话。 那么我的理想又是什么？ 我还不知道，首先还是先让自己忙碌起来，因为我还没有到那种总是休息的年龄。 jlaky 发表于 2006-01-02 15:45:17 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 heroes -[ thinking ] “历史是科学家和哲学家永不气馁的对宇宙的探索，是艺术家以各种艺术形式对生活之美的表达，是诗人以细致的语言技巧对微妙感情的阐释，也是先知和圣徒的想象对高贵人性的挑战。”- Heroes of History 历史的英雄就是被历史铭记的人，他们都对世界的发展有推动作用。小时候就知道也梦想成为英雄，那时候的我心中的英雄是超自然的，幻想中的，小说中的，电视中的，电影中的。后来发现真正的英雄是真实的，就在我们生活的世界中。 翻阅了英雄的事迹，发现他们往往都有过困苦的经历，但是都有广阔的胸襟，专注于某个领域，他们都是愿意为理想而牺牲的人。 jlaky 发表于 2005-12-20 21:41:00 阅读全文 | 评论(0) | 引用Trackback(0) | 编辑 分页: 模板设计： Danger