KM7 Overcoming cultural barriers to sharing knowledge

PART I

1. Culture is often seen as the key inhibitor of effective knowledge sharing.
2. Potential users said that they liked share system online, but just didn't have time for it.
3. Sharing was not built into the culture enough for people to actually take the time to do it.
4. However strong your commitment and approach to knowledge management, your culture is stronger.
5. Companies that successfully implement knowledge management do not try to change their culture to fit their knowledge management approach.
6. We defined culture as the shared values, beliefs and practice of the people in the organization.
7. The main reason knowledge management progams fail is a lack of a clear connection with a business goal.
8. It is the most important for the style of your effort to match how things get done in your organization.
9. Link these invisible values and visible elements of knowledge management is the behavior of peer and managers.

PART II

Culture is often seen as the key inhibitor of effective knowledge sharing. This paper interview 5 companies to find their organization culture.

Culture of organization can split into two dimensions - visible and invisible. The visible one is like espouse values, philosophy or mission. Even the stories, space and sturcture of one company is also regard as visible culture. And the invisible cultures such as their simple precepts are seen but unspoken background of the company.

Although culture affect knowledge sharing deeply, companies need not to change their culture to fit knowledge approach. Because culture is the foundation of a company, it is stronger as well. Instead, try to make KM approachs to fit your own culture and existent network, in addition, to let your employee get used to them. It will exert knowledge management more successfully.

PART III

TOPIC: Organizational Culture

This paper didn't define the organizational cuture clearly. But after Dr.Pheobe's interpretion, I knew more about this paper.

We can observe Organizational Culture by three things, that are atifacts & behaviors, values and assumptions. These three things is like an onion's skins. From the outside to inside is as the order above. These three can be split into visible and invisible dimension as well.

PART IV

I think there is only one insight in this paper, that is, building KM on solving problems.

And in my opinion, that's OK even your culture need to fit KM approach, if you can really solve problems and get profit from KM. Because, as a whole, there are still number of companies do not have that kind culture which suit to fit KM. In this case, the company need to create a new one for KM approachs.

The conclusions were only made from the companies which have knowledge sharing already. This is the biggest blind part of this paper, the authors didn't considered about other companies which have not had KM yet.

There is another interesting thing, that is, don't let your employees feel you are undergoing KM activities. Make KM approach as a routine of their work, and encourage them to get used to. I agree with it, and I think if the employees know what you required is about KM at first, they will feel overloaded. But I still think it is need to let them know what is KM step by step. It may make them agree more what you did.

KM6 Assessing Knowledge Assets: A review of the models used to measure intellectual capital

PART I

1. Stewart defines intellectual capital as intellectual material - Knowledge, information, intellectual property and experience - that can be put to use to create wealth.
2. IC could be an addendum accompanying with traditional financial reports.
3. The value an organization place on its IC is wholly dependent upon the goals of the organization and the state of the market.
4. A 500-year-old system of accounting must make way for a system of non-financial knowledge flow and intangible assets that use new proxies.
5. The pursuit of measuring IC assets objetively is a noble but difficult one.
6. The measurement examples thus far have been too firm-specific and no set of indicators could hope to be general enough to encompass the needs of a variety of international and industry settings.
7. Pursuing standards at this point might be more harmful given the nascent stage of research development.

part II

This paper reviews six assessments of IC and interprets their strengths and weaknesses. In the content, the author mentioned that many companies agree IC is important, but a few companies really practice it.

According to this paper, IC have different definitions among scholars. But it still has some similar parts, for example, human capital, finance of firm, renewal and development, customer etc. IC assessment the paper listed are trying to measure value of above items, even it is intangible.

In conclusion, IC assesment is almost firm-based, but pursuing standards might be more harmful given the nascent stage of research development. Academic should keep push this field into advance.

PART III

TOPIC: Knowledge assets assesment

This paper consider knowledge assets as intellectual capital. We could notice that there are many ways to assess IC and each of them have strenghs and weakness. It is also firm-based.

PART IV

After reading this paper, I felt something seems to be lost. First, I still cannot practice IC assesment, it is too abstract. Second, IC assesments are usually customization, but the author didn't talk about how to choose or create an appropriate one for each company. Third, because of the firm-based assesment, how to judge the way you used is right and efficient? After all, right or wrong, good or bad, those concept is relative as well as basing on comparing. But firm-base assesments can't compare with each other.

Now that it cannot compare, and the items which be assessed resemble knowledge map. I think, somewhere, knowledge mapping is enough and can be replaced IC assesment with it.

[統計] 第七章 抽樣分佈與估計式

第一節 抽樣誤差

1. 不對母體抽樣的原因：

• 母體太大
• 無法知道母體的範圍
• 破壞性檢測

2. 要如何抽樣才能有效的推估母體? 在於估計誤差的大小。

3. 估計誤差來源：

• 抽樣誤差： 使用恰當的樣本數來估計母體
• 非抽樣誤差：樣本無代表性，計算錯誤等

第二節 抽樣方法

1. 簡單隨機抽樣
先將母體編號，再以抽籤方式抽出

2. 間隔抽樣
每隔幾個就取幾個

3. 分層抽樣
要先決定有哪些重要的"層"

4. 集群抽樣
抽樣前，先將母體分為好幾個相似的集群，再以集群為單位來抽樣。如從全國小學10000所裡抽20所。

5. 分段抽樣
使用複合式的抽樣方法。如先進行集群抽樣後，再由每個集群裡，簡單隨機抽樣。

6. 配額抽樣(主觀)
類似分層抽樣，但事先不知道母群的各層比例

7. 判斷抽樣
最為主觀無根據的抽樣方法。

[統計] 第六章 連續機率分佈

第一節 連續變項

1. 連續變項是連續的，如溫度、時間、身高、收入等，表示任兩個數值之間，會存在著第三個值。
2. 明天中午12點溫度為30度的機率為多少？這個問題是無意義的，機率為0。應該說溫度到達30度上下（+ - 5）的機率為多少，有一個區間才能表達連續變項的特性。
3. 機率密度並無多大意義，通常關心的是累積機率密度。

第二節 連續機率分佈

1. 均勻分佈

• 一連續變項x，其值介於a, b之間。假設每一點出現的機率都是均等，則此變數的機率分佈為連續均勻分佈。
• 均勻分佈的機率為圖形面積。
• 平均數 = (a+b)/2, 變異數 = (b-a)平方/12

2. 常態分佈

如果一個連續變項X，具有公式6.8的性質，就是常態分佈。

2.1 超幾何分佈、二項式分佈、波式分佈與常態分佈之間的關係

• 超幾何分佈的樣本數是母體的0.05以下的話，可用二項式分佈取代
• 在二項式中，若樣本n夠大，則可用常態分佈取代；即使樣本n不大，機率p接近0.5 也可。若樣本數大而機率小時，可用波式分佈取代。
• 當樣本到達無限大時，超幾何分佈、二項式分佈與波式分佈皆會變為常態分佈

3. 標準常態分佈(Z分佈)

• 在常態分佈的狀況下，另其平均數為0，標準差為1，就可轉換為標準常態分佈。
• Z分數 = (該變數-平均數)/變異數

4. 伽瑪分佈

• 如果一隨機連續變數擁有公式6.10的特性，就是伽瑪分佈。
• 平均數 = a*b, 變異數 = a*(b平方)
• 伽瑪分佈可用來計算等候時間。在波式歷程裡，單位時間成功次數為入，那麼等候第一個成功事件出現的時間，平均就要 b=1/入。若要等候至第n個成功(a=n)，等候的時間就是伽瑪分佈。
• = GAMMADIST(時間X, 第a次成功, 第一次成功時間b, TRUE)，TRUE為累積機率，FALSE為機率密度函數。

5. 指數分佈

• 為伽瑪分佈的特例，令 a=1, b=1/入 就是指數分佈。
• 平均數 = 1/入, 變異數 = 1/(入平方)
• 伽瑪分佈是在算第n次成功的等待時間，指數分佈是在算第一次成功的等待時間。
• = EXPONDIST(時間X, 第一次成功時間b, TRUE)。

6. 卡方分布

• 為伽瑪分佈的特例，令 a=v/2, b=2，就是卡方分布。其中 v(nu)為正整數的自由度
• 平均數 = ab = v, 變異數 = a(b平方) = 2v，與伽瑪分佈相同。
• 若一連續變項X為標準常態分布，則該變項的平方(X平方)是為自由度1的卡方分布。多個獨立變項 X1~Xn均是標準常態分布，則自由度可累加起來變成n。
• 自由度越大，卡方分布越接近常態分布。
• 卡方分布常用來檢定資料與模式的吻合度
• P(x>a) = CHIINV(P,v)，給正無限大累積到a點的機率，計算a點是多少。
• P(x>a)= CHIDIST(點a, v)，給a點，計算大於a點數值的累積機率。

7. F分布

• 兩變數U, V互相獨立，且均有卡方分布的性質，其自由度分別為V1和V2，則隨機變數X= (U/v1) / (V/v2)
• 平均數 = v2/(v2-2), 變異數 = P115頁
• F分布常用於檢驗兩變異數是否相等
• P(X>a)= FINV(P, V1, V2)，給正無限大到a點的機率，計算a點是多少
• P(X>a)= FDIST(a點, V1, V2)，給a點，計算大於a點的累積機率。
• 若要計算a點的累積機率1 - P(X < a)

8. t分布

• 若變數U和變數Z互相獨立，而U為自由度v的卡方分布，Z為標準常態分布，則變數X = Z/ [(U/v)根號]的分布就是t分布
• t分布平方 = F分布
• 當自由度很大時，曲線就會接近標準常態分布
• t分布和z分布常用於檢驗母體平均數
• P(a < color="#ff6600">兩尾端面積和, v)，給兩尾端面積和，求a點和b點(a, b兩點為正負關係)
• P(X <> a) = TDIST(點a, v, tails)，給點a，求大於a點的累積機率。若tails = 1 則可直接算出累積機率，tails = 2可算出兩尾端面積和。

※ 在Excel中關於卡方分布(CHIDIST, CHIINV)、F分布(FDIST, FINV)和t分布(TDIST, TINV)的函數，所算得的累積機率是由正無限大到該點的累積機率。而在常態分布函數的情況下(NORMDIST, NORMINV, NORMSDIST, NORMSINV)所計算出的累積機率，則是由負無限大開始累積。