這幾天連個網路 就會被同棟裡的某些白目
用 netcut 把連線切斷 而且切的頻率實在是太高
從早上切到晚 每換個 ip 就切 …

= = 硍 真是令人火大 ….
連我的人工動態浮動 ip 都沒有用了 …
netcut 偵測新 ip 的速度實在是太快 ….

於是 這招不行 就換別招 ….
來個動態浮動 MAC Addr. … 每隔個一陣子自動浮動 MAC位址
果然效果好很多 … 原因大概是 NET CUT 針對新的 MAC 位址不能及時反應XD

而且 一變動 MAC 就可以網路通暢無比 瞬間領導權就變在自己身上
所以 就這樣 你切我 我切你 切切切切切 ….. 無言
馬的 也有品一點麻 下載個 X 片 電影 卡X 不要在大樓裡抓麻 ~_~”


2PC protocal

在分散式的環境裡 如果要確保 Atomic Transaction 完整執行
則我們可以使用 2PC 協議來確保這樣的事情發生

一個 Atomic Transaction 可能分成很多個 sub-transactions
把各自的 sub-transaction 託付給相對應的機器去執行

接著發起者 會開始傳送訊息 問每一台機器 (受託者) : 你們能否接下責任
對於每壹台機器 IF yes : 回答 ok IF no : 回答 I can’t
( 所有的訊息 都會儲存在 local端 以 log的方式存檔 才以訊息的方式發送出去 )

之後發起者 衡量所有機器的意見
IF 全都 ok : 傳送 託付成功 給所有機器
IF 不全 ok : 傳送 別做啦 (因為有人不能做啦) 給所有機器

目標 :
不管在哪個時間點 有哪台機器壞掉
我們可以利用儲存在機器裡的 log 檔 修復整個動作
以確保 Atomic Transaction “完整執行” 或是 “不要執行”


整個暑假 也快告一段落
不然以我的個性 可能玩過去整個暑假
頭腦還是要持續運轉 才不會秀逗的 @_@
今天原本要南下 但是又因為一點事情 改成明天
年紀愈大 愈想做的事情 好像都不能如期達成
Anyway , 星期日就會在家裡了 …
好好的渡過暑假的 happy ending …
” 暑假 ” 這個名詞 也許也是我最後一次使用囉 XD

Decentralized Research Support System

現在的搜尋系統這麼發達 一個關鍵字可以跑出上千萬甚至是上億的結果給你
要怎樣從這茫茫大海中 尋找適合自己興趣的論文 還真的是要花費一般功夫
又 CiteSeer 與 ACM Digital Libraries 提供了很好的論文資料庫管理系統
儘管如此 我們花時間在找東西的過程 依然是很無聊又很瑣碎
未來如果有一個系統 他的架構是基於Multiagent system設計理念
每個使用者的電腦上 都會有一隻Agent
自動的去發現使用者的興趣 與最近的研究方向
這支Agent能夠和與自己的主人有類似興趣的使用者的 Agents
做互相溝通 互相推薦論文給彼此
此外 每隻Agent會定期的去找研究相關的資料庫(如Citeseer)
取得相關符合使用者興趣的論文 當然在首步的搜尋結果出來後
之後以推薦列表的方式 把每篇論文依照推薦程度排序 列給使用者觀看即可

分散式系統的好處是 資訊是散佈在每台電腦中
每台電腦可以當Server 亦可以當作Client
當我們要做檔案分享時 可以確定我們所存取到的檔案是最新的
相較之下 集中式系統的缺點就是在於我們都把檔案放在Server下
當Server掛點 整個系統就會損毀 因此Server的負載是很高的
而外 Server 的檔案可能並不是最新的 因為使用者可能忘記上傳
或是沒有上傳最新的文件到伺服器上 ….

繁瑣重複性的工作 我們交給電腦做就好
但是電腦要夠聰明才可 : )


我竟然看起華視偶像劇了 ~_~”
一看就無法自拔  一口氣看到結束
劇情還滿感人的  也滿搞笑的
詩詩裝淑女起來還真的不錯看 XD
一個重考班能搞出這麼多名堂  還真不是蓋的啊
( 比較像把咩教學班 …. )

Progress report

Progress report

By Yu-Chung Shen

All that I do in the summer vacation can be listed as the follows:

About research:

1. (Date: 07/10) I presented a paper about recommendation system in CS group which use the popularity measurements to characterize products. Based on this method, each product can be categorized into a specific popularity class and then we can identify each target customer with their favorite class, finally recommend products of the class to them.

2. (Date: 08/07) Again, I presented a paper about RS in CS group. This paper proposes a multi-agent system called IMPLICIT, which can recommend us with the web links when we use search engines. The design goal of the system is to help us reduce the time of finding the items we want in the search result list. We can image that the keywords we query might have been queried before by other people. So we can design agents for each user which can cache items of search result the user like. Next time other users query the same keyword we can directly recommend the search result which the previous users like and who have queried it before.

3.(Date: 07/14) I meet 左哥 at Lab to discuss the work about research successor. I found that 左哥 is an ambitious guy and have the dream to develop a system that can automatically extract the metadata of every object in this world. He is also a web technology fanatic since he had told me and analyzed for me the difference between many technologies nowadays. From his previous work, which is a system help user manage their paper readings online, I learn the implementation detail such as OpenID, DHT, RoR, and the architecture of REST design protocol. Even thought he is busy everyday doing his work at company, we sometimes talk via email, I think I still have lots more to work on and 左哥 can be my good advisor.

4. I surveyed some papers about research support system that can help researcher to do their research. The vision of this type of system is to provide an environment that allows users to manage their readings or publications and create a research group of a specific research topic. When the system have cached enough data (readings or publications), the system can automatically identify each user or group with their interests and recommend items (readings or publications) concerning about their interests to them.


1. I regularly attend OS and Complexity study group with 文杰 each week. These groups are founded for discussing subjects about qualify exams for PHD students. My goal is to pass at least 3 subjects in the exams (2007 Fall) ,and I deeply hope 文杰 can successfully pass all the subjects in the exams. Since he had taught me lots of things that I didn’t know before, and refresh me with his special thought or ideas about research in the AI domain. We deeply hope that we can collaborate on doing research or to attend conference together in the future days. We believe that even though we are not genius but the power of the collaboration can be respectable if we all trust our abilities.

2. I’ve been at home (Kaohsiung) for 11 days, been at Taichung for 6 days. During these days, I visit my dear parents, my friend, my puppy, etc. I just take a rest and refresh my mind.