Knowledge Management for Reuse

Please download to get full document.

View again

of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Knowledge Management for Reuse
Document Share
Document Tags
Document Transcript
  Conference Proceedings – distributing knowledge in buildingInternational Council for Research and Innovation in Building and ConstructionCIB w78 conference 2002Aarhus School of Architecture, 12 – 14 June 20021Theme:Title: Knowledge Management for Reuse Author(s):Renate Fruchter, Peter DemianInstitution(s):Department of Civil and Environmental Engineering, Stanford UniversityE-mail(s), pdemian@stanford.eduAbstract:  Managing and reusing knowledge in architecture, engineering and construction firms can lead to greater competitive advantage, improved designs, and moreeffective management of constructed facilities. We define design knowledge reuseas the reuse of previously designed buildings, building subsystems, or buildingcomponents, as well as the knowledge and expertise ingrained in these previousdesigns. This paper introduces the notion of knowledge in context . We argue that in order for knowledge to be reusable, the user should be able to see the context inwhich this knowledge was srcinally created and interact with this rich content.We call a repository of such knowledge in context the corporate memory. Wedescribe empirical observations of designers reusing knowledge from their  personal design experiences. Based on these observations, we formalize two keyactivities in the process of knowledge reuse: finding  reusable items and  understanding  these items in context. We formalize six degrees of exploration that lead to understanding. We describe a prototype knowledge management system,CoMem (Corporate Memory), that supports these activities. CoMem isdistinguished from the document-centric state-of-practice solutions by its approachof “overview first, zoom and filter, and then details-on-demand.” Keywords:  Design reuse, corporate memory, human-computer interaction, knowledgemanagement, knowledge reuse Introduction The average designer, whether consciously or subconsciously, draws from a vast well of previous designexperience. This can be experience acquired by the individual or by his/her mentors or professionalcommunity. We refer to this activity as design knowledge reuse . Specifically, we define designknowledge reuse as the reuse of previously designed artefacts or components, as well as the knowledgeand expertise ingrained in these previous designs. We distinguish between two types of reuse:1.    Internal knowledge reuse : a designer reusing knowledge from his/her own personal experiences(internal memory). For example, a structural designer might remember that the last time shedesigned a floor slab for a hotel ballroom it was too thin, which resulted in vibration problems. Thenext time she is faced with a similar design situation, she designs the floor slab deeper.2.    External knowledge reuse : a designer reusing knowledge from an external knowledge repository(external memory). For example, the same structural designer might look for floor slab designs inher company’s standard components database. She retrieves a floor slab design that comes with aspreadsheet for calculating the correct slab thickness. This spreadsheet takes into account thecompany’s previous experiences with bouncy floor slabs and increases the depth beyond theminimum required by the building code.Whereas internal knowledge reuse is effective, external knowledge reuse often fails. This failure occursfor numerous reasons, including: •   Designers do not appreciate the importance of knowledge capture because of the additional overheadrequired to document their process and rationale. Consequently, knowledge is often not captured. •   Even when knowledge capture does take place, it is limited to formal knowledge (e.g. documents).Contextual or informal knowledge, such as the rationale behind design decisions, or the interactionbetween team members on a design team, is often lost, rendering the captured knowledge notreusable, as is often the case in current industry documentation practices. •   There are no mechanisms from both the information technology and organizational viewpoints forcapturing, finding, and retrieving reusable knowledge. Construction Informatics DigitalLibrary paper w78-2002-20.content  Conference Proceedings – distributing knowledge in buildingInternational Council for Research and Innovation in Building and ConstructionCIB w78 conference 2002Aarhus School of Architecture, 12 – 14 June 20022 Empirical observations of designers at work show that internal knowledge reuse is effective since: •   The designer can quickly  find   (mentally) reusable items. •   The designer can remember the context of each item, and can therefore understand   it and reuse moreeffectively.We use these observations of internal knowledge reuse  to improve external knowledge reuse .We introduce the notion of knowledge in context  . Knowledge in context is design knowledge as it occursin a designer’s personal memory: rich, detailed, and contextual. This context includes design evolution(from sketches and back-of-the-envelope calculations to detailed 3D CAD, analysis, and simulations),design rationale, and relationships between different perspectives within cross-disciplinary design teams.We define the corporate memory  as a repository of knowledge in context; in other words, it is an externalknowledge repository containing the corporation’s past projects that attempts to emulate thecharacteristics of an internal memory, i.e. rich, detailed, and contextual. The corporate memory grows asthe design firm works on more projects.We view knowledge reuse as astep in the knowledge life cycle(Figure 1). Knowledge is createdas designers collaborate on designprojects. It is captured, indexed,and stored in an archive. At alater time, it is retrieved from thearchive and reused. Finally, asknowledge is reused it is refined and becomes more valuable. In this sense, the archive system acts as aknowledge refinery. The current study focuses on the knowledge reuse phase and builds on previouswork that addresses knowledge creation, capture, indexing, and archiving (Fruchter 1996, Fruchter et. al.1998, Reiner and Fruchter 2000).The practical motivation behind the development of external knowledge reuse systems is that the captureand reuse of knowledge is less costly than its recreation. In current practice knowledge capture and reuseare limited to dealing with paper archives. Even when the archives are digital, they are usually in theform of electronic files (documents) arranged in folders which are difficult to explore and navigate. Atypical query might be, “how did we design previous cooling tower support structures in hotel buildingprojects?” In many cases, the user of such systems is overloaded with information, but with very littlecontext to help him/her decide if and what to reuse.This paper addresses the following central questions: •   What are the key characteristics of the knowledge reuse process? •   How can the reuse process in the construction industry be supported by a computer system? •   What are natural idioms that can be modelled into a computer system to provide an effectiveknowledge reuse experience to a designer?Our objective is to assist the designer and to support the reuse process rather than to automate it. Related Research Related research studies on design knowledge reuse focus either on the cognitive aspects or on thecomputational aspects.Research into the cognitive aspects of reuse has helped to identify the information needed by designers.For example, Kuffner and Ullman (1990) found that mechanical engineers usually request informationconcerning the operation or purpose of a designed object, information that is not typically captured instandard design documents: drawings and specifications. This research extends these findings byformalizing the requirements for contextual information when reusing items from previous projects.On the computational side, research into design knowledge reuse focuses on design knowledge representation  and reasoning . Knowledge representation ranges from informal classification systems forstandard components to more structured design rationale approaches (Hu et. al. 2000 gives an overview).There is a trade-off in design rationale systems between the overhead for recording design activities andthe structure of the knowledge captured. Knowledge creationKnowledge reuseARCHIVE CaptureIndexStore Knowledge refinement ? FOCUS OF THIS RESEARCHLEGEND Figure 1: The knowledge life cycle  Conference Proceedings – distributing knowledge in buildingInternational Council for Research and Innovation in Building and ConstructionCIB w78 conference 2002Aarhus School of Architecture, 12 – 14 June 20023 Highly structured representations of design knowledge can be used for reasoning . However, theseapproaches usually require manual pre or post processing, structuring and indexing of design knowledge.For example, ARCHIE is a case-based reasoning tool for aiding architects during conceptual design(Domeshek and Kolodner 1993). CASECAD enables designers to retrieve previous design cases basedon formal specifications of new design problems (Maher 1997). Such tools enable knowledge reusebased on a priori set representations that are specific to narrowly defined domains and media types.This research brings together the cognitive and computational approaches. We consider reuse to be acombined effort involving both the human and the computer. We therefore address the issue of designknowledge reuse as a human-computer interaction (HCI) problem, and we take a user-centred approach todesigning this interaction. We aim to provide a knowledge reuse experience that leverages natural idiomsand metaphors in order to support the designer in doing his/her work, and we consider automaticreasoning approaches to constrain the user’s knowledge reuse activities.In our approach, capture and indexing take place in real time, with the least possible intrusion on thedesign process. We take as our point of departure the Project Memory (ProMem) system (Fruchter et. al.1998, Reiner and Fruchter 2000), which transparently captures the evolution of building design projectsby supporting the typical communication and coordination activities that occur during collaborativedesign. ProMem is based on the Semantic Modelling Engine (SME) (Fruchter 1996), which is aframework that enables designers to map objects from a shared graphic product model to multiplesemantic representations and to other shared project knowledge. Scenario-based Design of a Corporate Memory System CoMem (Corporate Memory) is a prototype system that extends ProMem firstly by grouping theaccumulated set of project memories into a corporate memory, and secondly by supporting externalknowledge reuse from this corporate memory.In developing CoMem, we adopted a scenario-based approach to the design of human-computerinteraction (Rosson and Carroll 2001). The premise behind this method is that descriptions of peopleusing technology are essential in analysing how technology can improve or support their activities.The scenario-based design process begins with an analysis of current practice using  problem scenarios .These are transformed into activity scenarios , which are narratives of typical services that users will seek from the system.  Information scenarios  are elaborations of the activity scenarios which provide details of the information that the system will provide to the user. Interaction scenarios describe the details of userinteraction and feedback. The final stage is  prototyping  and evaluation based on the interaction scenarios.The process as a whole from problem scenarios to prototype development is iterative.CoMem is being developed using three sets of scenarios. In the first scenario, a novice designer usesCoMem to find and reuse a design component about which she knows very little. In the second scenario,an expert and a novice designer use CoMem together, with the expert using CoMem as a mentoring tool.Finally, in the third scenario, an expert designer uses CoMem to retrieve a design component about whichshe already processes a lot of contextual information, and so she uses CoMem simply as a retrieval tool.Ongoing development of CoMem is guided by the iterative analysis, testing, and refinement of thesescenarios. This paper focuses on the first scenario, that of a novice using CoMem to find and understanda component about which she knows very little. A Design Knowledge Reuse Scenario As part of our analysis of current practice in order to generate and analyse realistic problem scenarios, weconducted a two-week ethnographic study in a structural design office in California. The objective of thestudy was to investigate the reuse process qualitatively, and gain a deeper understanding of the stepsinvolved and the types of information reused. Data was collected by videotaping, transcribing, andcoding design meetings and site visits. Internal knowledge reuse was observed and recorded when anovice structural designer asked an expert designer questions. Our observations indicated that the expertalways referred to his work on previous projects when answering these questions.The following is an example of a problem scenario that was developed based on our ethnographic study.  An expert structural designer, Eric, and a novice, Nicola, both work for a structural design office inCalifornia. The office is part of the “X Inc” Structural Engineering Firm. They are working on a ten-story hotel that has a large cooling tower unit. Nicola must design the frame that will support this  Conference Proceedings – distributing knowledge in buildingInternational Council for Research and Innovation in Building and ConstructionCIB w78 conference 2002Aarhus School of Architecture, 12 – 14 June 20024 cooling tower. Nicola gets stuck and asks Eric for advice. Eric recalls several other hotel projects that were designed by “X Inc”. He tells Nicola to look at the drawings from the Bay Saint Louis project, ahotel project that “X Inc” designed a couple of years ago (Figure 2). Nicola spends over an hour looking for the Bay Saint Louis drawings in the “X Inc” paper archive. Sheeventually finds the drawing sheet with the Bay Saint Louis cooling tower frame. She shows it to Eric.The drawing shows the cooling tower frame as it was finally built. It is a steel frame. Eric realizes that what he had in mind for Nicola to reuse is an earlier version that had a steel part and a concrete part. He is not sure if this earlier version is documented somewhere in the archive. Rather than go through the paper archive again, Eric simply sketches the design for Nicola. Eric’s sketch also shows the load pathconcept much more clearly than theCAD drawing would have, which helps Nicola to understand the design. Ericexplains to Nicola how and why thedesign evolved. Given the current  project they are working on, it would bemore appropriate to reuse the earlier composite version. Eric recalls that thespecifications of the cooling tower unit itself, which were provided by the HVAC (heating, ventilation and air conditioning) subcontractor, had a largeimpact on the design. Nicola now feelsconfident enough to design the newcooling tower frame by reusing the sameconcepts as the Bay Saint Louis cooling tower frame, as well as some of the standard details. Using scenarios such as the one above, we observed that the expert was always able to find relevant itemsto reuse from his personal experiences, and he always described the project context and evolution of eachitem to help the novice understand it (Figure 2). Formalized Process of Design Knowledge Reuse During our study, the expert’s internal knowledge reuse process was observed to be very effective. Hewas always able to recall directly related past experiences and apply them to the situation at hand. Twokey observations in particular characterize the effectiveness of his internal knowledge reuse:1.   He was always able to  find   relevant designs or experiences to reuse.2.   For each specific design or part of a design he was reusing, he was able to retrieve a lot of contextualknowledge. This helped him to understand   this design and apply it to the situation at hand. Whendescribing contextual knowledge to the novice, the expert explored two contextual dimensions: the  project context   and the evolution history .The  project context   dimension encapsulates the levels of granularity at which contextual knowledge aboutthe design project can be explored. Given an item from the corporate memory, we identified thefollowing directions of exploration: •   UP: From component to subassembly . Designers move upwards along this dimension to explorethe discipline and project in which this item occurs. For example, if a structural designer considersreusing a cooling tower frame from a previous project, he/she might recall the structural system orthe entire project from which this cooling tower frame is taken. •   DOWN: From subassembly to component . Designers move downwards along this dimension toconsider the subparts or subcomponents of which this item is composed. For the cooling towerframe, the structural designer can consider the individual beams, columns, braces, and connections of which the frame is composed. •   SIDEWAYS: From one item to related items . Designers move sideways to explore related itemsin the corporate memory. For the cooling tower frame, the designer reusing the frame can consider Figure 2: Reuse scenario  Conference Proceedings – distributing knowledge in buildingInternational Council for Research and Innovation in Building and ConstructionCIB w78 conference 2002Aarhus School of Architecture, 12 – 14 June 20025 the cooling tower unit supported by the frame to determine what load it exerts on the frame, or he/shemight explore architectural features related to the frame.The evolution history  is the record of how an item evolved from an abstract idea or a set of requirementsto a fully designed physical entity. Given an item from the corporate memory, we identified thefollowing directions of exploration: •   UP: From detailed to conceptual . Designers move upwards along this dimension to trace theconcepts that were explored early on in the design of this item. For the cooling tower frame, thedesigner might recall a sketch of the conceptual braced frame design that was created early in thedesign process. •   DOWN: From conceptual to detailed . Designers move downwards along this dimension to followthe evolution of this item into a fully designed physical component. For the cooling tower frame, thedesigner reusing the frame might follow its evolution into a fully detailed design in a CAD file, andcan even study photographs of the frame as built. •   SIDEWAYS: From alternative to alternative . Designers also move sideways to explore thedifferent alternatives that were considered at any stage in the design process. The designer reusingthe cooling tower frame might recall that steel and concrete alternatives were considered. Perhapsthe concrete alternative that was srcinally abandoned can now be reused.There are therefore six degrees of exploration , three – up, down and sideways – in each of the twocontextual dimensions.The observed process of internal knowledge reuse is formalized into three steps (Figure 2):1.   Finding a reusable item2.   Exploring its project context in order to understand it and assess its reusability3.   Exploring its evolution history in order to understand it and assess its reusabilityWe use these observations of internal  knowledge reuse as the basis for supporting external knowledgereuse from a corporate memory . The CoMem Approach CoMem is designed to support the same activities observed during the expert designer’s internalknowledge reuse process. The CoMem human-computer interaction experience is based on the principleof “ overview first, zoom and filter, and then details-on-demand  ” (Shneiderman 1999). Based on the threereuse activities identified above – find, explore project context, explore evolution history – CoMem hasthree corresponding modules: an overview , a project context explorer  , and an evolution history explorer  .The overview  supports the designer in finding reusable items. The objective is to enable the designer toview the entire corporate memory at a glance. The overview gives the designer an indication of which“regions” of the corporate memory contain potentially reusable items. The overview might be extremelydense. Filtering tools are used to avert information overload and help the designer focus by addingemphasis to certain items.Once the user has selected an item from the overview, the project context explorer  supports the designerin exploring this item’s project context. This shows the project and discipline to which this item belongs,as well as related components, disciplines and projects. The item selected from the overview becomes thefocal point of the project context explorer.Finally, in the evolution history explorer , the designer can explore the evolution history of any itemselected from the overview. This view tells the story of how this item evolved from an abstract idea to afully designed and detailed physical artefact or component. CoMem Overview Module  The interaction design process for this module is based on the empirical observation that a designer canfind reusable items from his/her internal memory. We argue that an external memory system (CoMem)needs to support the finding activity. We translate this activity need into an information need: the needfor an overview of the corporate memory.
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks