Table of Contents

Cognitive Systems Engineering

0471011983.jpg

Citation - Rasmussen, J., Pejtersen, A. M., & Goodstein, L. P. (1994). Cognitive systems engineering. New York: John Wiley & Sons, Inc.

Keyword - cognitive work analysis, System Engineering, Information Science, Information Behavior

Introduction

Analysis

Work Domain Analysis / 工作領域分析

Delimitation of the work system / 界定工作系統

Levels of abstraction: the Means-ends relations / 概念提取層次:手段-目的關係

General Work Activities & Functions
Physical Processes
Abstract Functions and Priorities
Interrelationship between levels
Goals and Constraints
Physical Configuration and Anatomy

Thus the verbal labels identifying the tools and equipment at this level refer normally to the processes of function being served at higher levels of the mean-ends network. For instance, consider the terms “typewriter,” “pump,” and “screw driver”. In addition to verbal descriptions, this level also includes representations in the form of pictures, topographic maps, drawings of buildings and equipment, and so on.

Fig 2.4. WHY, WHAT, and HOW in the mean-ends space

The Structure of the Means-Ends Network / 手段-目的網路的結構

如上所述,手段-目的結構將極端的系統整體目標與物質資源使用結合起來。

由以上對手段-目的層次的描述,可以看作是由一個概念層次到下一個概念層次,表現其功能元素的概念與反映其功能結構形式的轉移。 / From the above description of the means-ends levels, it is seen that a shift from one level of abstraction to the next involves a change in both the concepts used to represent the functional elements and in the format used to reflect the functional structure.

不同層次所包含的資訊都是與相同的現實世界相關。而呈現不同層次所用的資訊,是用來作為一組物質資源世界與終極目標與限制的概念性轉換連結。

For example, when moving form one level of abstraction to the next higher level, the change in the represented system properties is not merely a removal of details of information about the original level. More fundamentally, information is added on higher level principles governing the cofunctioning of the various elements at the lower level. If traced to the top level, it is clear these higher level principles derive originally form the overall purpose of the system (i.e., form the reasons and intentions behind the design).

工作領域的手段-目的關係透過此處的描繪呈現外顯化,常見於工作系統的系統分析工作。在技術領域的分析中最為清楚。

Levels of Decomposition: Part-Whole Relations / 分解層次: 部份-整體關係

Choosing the proper levels of decomposition as well as representation of the productive, functional resources is rather straightforward since this is related to the usual methods of functional analysis. In addition, as discussed in the introduction, categorical representations are adequate for feedback system designs.

Coupling to the Environment / 結合環境

Coupling to the Actor: Categories of Work System / 結合行動者: 工作系統的歸類

To ensure effective information retrieval support consistent with the above, it is necessary to make information about work domain properties accessible to decision makers from several perspectives. Thus answers will be required to queries such as: “what” is available, “how” can this be accomplished, “why” is this required? This means that the search term are not objective defined by the collection of stored items but by the means-ends relations among them. As discussed in the previous section, the representation of the productive, functional features at each level should be comparable to that of the intentional aspects in order to facilitate decision making and choosing among options.

The Means-Ends Network as Behavior Shaping Constraints / 手段-目的網路是形塑行為的限制

Source of Regularity of Behavior of Work Domain / 工作領域行為的規則性來源

透過工作領域中內部意圖限制與功能性因果限制的比較,可以了解不同工作領域規則性來源的特性。在此一規則性來源的光譜中,一端是緊密連結、技術系統、其規則性來自於穩定的自然規律;另一端是取決於使用者的主觀特性與價值觀。而這兩種極端中間,有許多不同的系統,是兩者兼具的。

行動者意圖式規則結構 ——— 自然規律式環境規則結構
自然環境
鬆散的元素物件組合
包含工具與人工製品的環境
低警覺
可逆
允許嘗試與錯誤
緊密結合高度結構化系統
高度警覺
潛在不可逆
不接受嘗試與錯誤

相對的現實系統

一般公共資訊系統 公共服務系統
圖書館
人工生產系統
醫院
機械化生產工廠 自動化生產製程
自然法則(自然律)主導的自動化系統

生產指導規則主導的製造系統

行動者意圖主導的鬆散結合系統

(pp. 51-52 )

在各種鬆散結合的工作領域中,可以發現一種Just-in-time工作方法取向,例如醫院、辦公室、人工為主的製造工廠。在這種系統中,生產性功能並非由技術核心限制,而是在於員工本身的活動。也就是說,當使用者集中注意的時候,並不存在一種固定的實體系統作為使用者介面。

行動的協調與控制乃是根據對組織目標的溝通而來。

行動者個人主觀主導的系統

比光譜更左側(行動者意圖性結構)的系統。出現在研究機構與大學,與第九章描繪的公共圖書館。

然而,系統設計受助於能被使用者需求所發現的典型的「工作環境」,並且其工作實踐與意圖能為實地研究所發現。 / However, system design is facilitated considerably by the fact that typical “work situations” can be identified in terms of user problems and needs, and their work practices and intentions can be identified by field studies.

獨立自主系統,因果的使用者

某些一般公用資訊服務資訊系統,如詞彙編撰資料庫等,並不清楚其使用者的任務情境與意圖。這些系統通常涵蓋了一些普遍的有用資訊來源,並且資料庫是由許多不同的公用服務、機構、與商業公司所產生的。沒有一通用的相關性意圖結構或任務情境,以形成設計系統基本功能架構與設計介面。任務情境的意圖性完全由實際需要的使用者所定義。除了以一般性的資訊以外,沒有更高層次的手段-目的結構。

系統分類的結論與評論

(pp. 52-53) 這五類系統的是根據其意圖性的屬性而加以歸類。其意圖性結構乃根據實際使用者主觀依賴的法則(laws)、常規(regulations)、社會規則與價值(social rules and values)所決定。

對處理複雜系統的行動者而言,控制事物的狀態需要考慮各種功能性與意圖性關係的組合。

因此,如果工作系統的互動會透過資訊系統協調,則此系統的意圖性結構的限制與功能性因果關係的限制都必須描繪出來。 / when the interaction with a work system is mediated through an information system, constraints originating in the intentional structures, as well as those based on functional, causal relations, must be represented.

此一工作領域描繪的分層結構,在管理與組織理論中也有相關類似的概念 (Parsons, 1960; Thompson, 1967; 見第四章)。任何管理決策或組織模型建立,都不能不考慮所有的手段-目的層次與系統基本目的。

最近 Mitroff and Linstone (1993)的組織模型專書中,參考了 Churchman (1971) 的分類。此書主題是討論在高度整合與動態的社會技術系統中的決策問題。(以下段落討論此書與本書論點,略)

Aristotle 四因說與means-ends relations
物質因
material cause
形式因
formal cause
效力因
efficient cause
目的因
final cause
What is it made of? What is it? How was it made? Why was it made?
WHAT - physical WHAT - conceptual means (HOW) ends(WHY)

Hints For Field Analysis / 實地分析注意事項

工作領域分析的資料有許多來源。在開始實地研究之前,可以由教科書、活動報告、機構簡介等等,獲得該領域的基本介紹。新聞與報紙可能會提供相關政策、管理方向、市場環境等資料。在更細節的部分,正式的程序、工作指導說明,能夠提供系統相關資訊。

Interview techniques

Activity Analysis / 活動分析

工作領域活動 Activity in work domain terms

原型活動元素 / prototypical activity elements
原型工作處境 / prototypical work situation
Example

Activity analysis example (1)

工作處境間的溝通 / communication among work situation

連結工作處境與行動者的溝通結構 是 設計整合資訊系統的重要面向。 醫院與製造業的溝通結構相當不同。 / The communication structure connecting work situation and the actors is an important aspect for the design of integrated information systems. This structure is very different in the hospital and manufacturing cases.

在醫院的原型處境中,行為者的工作團隊是由專業角色所組成的,而非個人特質。 而為了轉移工作需求,大多數醫院中的工作角色不是指派給個人。 因此,工作處境與角色之間有穩定的設定。 / In the prototypical work situations of the hospital, several actors work as a team whose members are identified with reference to roles and professional background, not to persons. Due to shift-work requirements, most roles in the work of a hospital are not assigned to particular individuals. Therefore, a stable configuration is linked to work situations and roles.

製造業並不會轉移工作,角色也與特定個人有關。原型工作功能並非以集中在特定的時間處境,而是跨時間的處境。 活動的固定設定與特定個人以及其工作功能有關。 / In the manufacturing, no shift work was involved and the roles were closely related to particular persons. The prototypical work functions were not concentrated in time to particular situations, but spread over time. A stable configuration of activities was related to particular individuals and their work functions.

[how to analysis]

決策活動 Activity in decision making terms

[how to analysis] 決策活動分析將描繪在各種原型任務中發現的決策功能,以關聯各種反覆出現的「知識型態」。行動者藉由「知識型態」表徵工作領域中的事物狀態,並且透過計劃中的元素控制事物狀態。 / This part of the analysis will describe the decision functions found in the various prototypical task situations and functions in order to interrelate the various recurrent “state of knowledge” used by the actors to represent the state of affairs in the work domain and the elements of their plans to control such states.

[control functions] Several different categories of information processing are necessary for these control functions.

  1. 透過分析確認環境實際狀態:資訊收集、處境分析、診斷 / The actual state of the environment must be identified; this involves information collection, situation analysis, and diagnosis, which are all analytical process.
  2. 目的的評估參考:預估、價值判斷(評價)、挑選 / Next, the implications of the actual state with reference to the current objectives must be evaluated. This process involves prediction, value judgment, and choice.
  3. 選定適當的控制工作次序:計劃、時間安排 / Finally, the proper sequence of control actions must be selected by a process of planning and scheduling.
structure of decision sequence

將以上控制程序標準化,建立一通用的決策結構,為決策梯(decision ladder)。此決策梯根據不同工作領域的實際工作的口語分析(verbal protocol analysis)[即think-aloud],找出行動者對其知識行動的歸類,與這些決策知識行動在時間上的順序。

決策梯。Fig.3.6

[圖解]在此決策梯中,有兩種不同的行動順序。一種是基準化的、合理的決策順序(normative, rational sequence),另一種是自然而然,非正規的啟發式的捷徑(heuristics shortcut)。箭頭表示規範式的決策行動順序,但是也常常發現有相反方向的行動順序。在實地觀察中,此基準架構像是一種「記事本」,用於紀錄與呈現所觀察到的決策行為、決策路徑、與資訊溝通。[像是編碼簿,幫助研究者將觀察到的實地行為編碼概念化]

Formal decision making

在新手,或是專家面臨不熟悉的處境時,會採用基本的基準化的決策順序,而沒有任何啟發式捷徑。 It depends on knowledge about the internal, functional, and international properties of the work system.

Subroutines connect different states of knowledge about the state of affairs in the work system and this, according to the discussion in Ch2, can be done in terms of various levels of abstraction and decomposition. Information processing then involves the transformation of one state of knowledge to another.

The transformations are based on mental models representing the constraints of the work environment. These comprise the invariant properties that govern its normal regularity of behavior and that determine the internal relationships in the set of data provided by the environment.

Heuristic decision making

有技巧的行動者藉由其「know-how」知識,能以「啟發式模式」作為基本的決策程序。

人並非被動的接受資訊,而是在活動中持續不斷的提出對環境的問題。設想,人具有高度的調適性,並且傾向減低其認知壓力(以最小努力原則),因此熟手只有注意部分必要的行動線索。這些「know-how」是由經驗而來。

啟發式決策活動的形成 是 一種由環境訊號線索而來的持續性控制活動。此觀點接近Gibson 的 direct perception of affordances 的主張。也近似於 Hammond 的「準理性思維」理論中,對理性與直覺判斷的區分。

資訊程序策略活動 Activity in Info Processing Strategies terms

分析行動者應用不同決策任務時的心智策略特徵。

定義心智策略(mental strategies)

策略是一種認知程序。 A “strategy” is taken here to be a idealized category of cognitive process, which can serve one the the decision processes discussed in the previous section.

策略以不同形式落實在工作處境中,但是有共同的特徵,如特定的心智模式、解釋可見證據的模式、與相同的策略計劃規則。 / The actual implementations of a given strategy during a work situation will be different each time, but they will share important characteristics, such as a particular kind of mental model, a certain mode of interpretation of the observed evidence, and a coherent set of tactical planning rules. (p. 70)

根據定義,不同的策略需要不同的行動者資源設定:知識、認知處理能力、時間,等等。因此策略也是一種認知程序的抽象化概念結構。

範例
策略類別 展開(所根據的資訊) 資訊程序
歸納 情境經驗庫 受到有限行動選項影響的辨識 技術診斷中的辨識案例
假設-演繹 功能性結構 由對原因的假設演繹症狀 技術與醫療的分析性診斷
演繹並搜尋 問題空間的理性結構 演繹,在解答空間中檢索 圖書館中的分析搜尋
搜尋並比較 搜尋實體或功能性地貌 檢索並以樣版比較 技術性診斷的地貌式檢索;瀏覽圖書館
同上 搜尋決策樹 同上 技術與醫療診斷的決策表格檢索;圖書館中的經驗策略
Implications for system design

活動分析的方法論技巧 Methodological hints for Activity analysis

找出原型任務情境
工作領域活動分析
決策活動分析
心智分析:行為軌跡分析

確認出有意義的決策功能之後,需要分析行為軌跡,以分解並歸類心智活動,發現決策功能中所使用的各種資訊程序策略。

分析的原則 行為片段須被個別分析;行為片段須由熟悉工作領域的人在情境中所了解。

工作的完成是由個人主觀標準所主導;因此,在理解、結構化、歸類資料之前,進行樣本的平均會導致誤解。

事件軌跡分析可參見 Hollnagel et al.(1981);行為軌跡分析可參見 Sanderson and Fisher (1994)。

各種分析階段:

  1. 原始資料 / Raw Date: 由訪談、觀察、實地成效測量、模擬設備所取得的研究資料。
  2. 中介資料 / Intermediate Data: 由原始資料進行第一階段處理。合併資料並以時間排序,以描述其中的具體特定「事件(occurrence)」。此描述需要以實地原本的詞彙。
  3. 已分析之事件資料 / Analyzed Event Data: 以正規化詞彙與概念描述工作任務,以反映出分析的理論基礎。
  4. 概念描述 / Conceptual Description: 以類別描述常見的、統一的事件片段的普遍特性。
  5. 能力描述 / Competence Description: 描述行為軌跡中的認知控制。

範例:以放聲思考法分析心智策略

Analysis of Work Organization and System Users

User-Work Coupling

At the Periphery of Effective Coupling: Human Error

Application

The Design Process and Its Guidance

Evaluation of Design Concepts and Products

Example

Design of a Library System

BookHouse Design: Data Base and User Dialogue

BookHouse Design: Interface Displays

BookHouse Evaluation

Catalog of Annotated Displays

Metadata

Author - Rasmussen, J., Pejtersen, Goodstein

Tags -