Presentation is loading. Please wait.

Presentation is loading. Please wait.

Designing and validating a structuring method for a multipurpose knowledge repository And evaluating the method in a software prototype Lars van der Meer.

Similar presentations


Presentation on theme: "Designing and validating a structuring method for a multipurpose knowledge repository And evaluating the method in a software prototype Lars van der Meer."— Presentation transcript:

1 Designing and validating a structuring method for a multipurpose knowledge repository And evaluating the method in a software prototype Lars van der Meer Universiteit Utrecht Master Business Informatics 28-08-08 1/53

2 Agenda Background Awareness of problem Research question Research method Results Conclusions Discussion and future work 2/53

3 Background 3/53 Knowledge Knowledge management Knowledge repository Multipurpose knowledge repository General Specific Knowledge Data: ‘Known facts or things used as a basis of inference or reckoning’ Information: ‘Systematically organized data that tells us something when we interpret it’ Knowledge: ‘Actionable information’ (Jashapara, 2004) Knowledge management ‘..all methods, instruments and tools that in a holistic approach contribute to the promotion of core knowledge processes’ (Mertins et al., 2000) Knowledge repository ‘A knowledge repository is a tool that contributes to knowledge management by supporting capturing, sharing and (re)usage of knowledge’ Multipurpose knowledge repository ‘A knowledge repository that enables different organizations to structure knowledge in a way that fits their knowledge domain and purpose of use’

4 Background ‘It has become all too easy to accumulate knowledge in digitized form … after a point, search and recontextualization costs outweigh the potential benefits from reusing the knowledge’ (Raghu, 2005). Challenge is making knowledge efficiently retrievable Search engine Browsing 4/53

5 Agenda Background Awareness of problem Research question Research method Results Conclusions Discussion and future work 5/53

6 Awareness of problem Domain modelling schemes o Taxonomy 6/53

7 Taxonomy Windows explorer Knowledge system (TOPdesk) 7/53

8 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus 8/53

9 Thesaurus Category FAQ NT FAQ Category BT Points of interest RT 9/53

10 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus o Ontology 10/53

11 Ontology A vocabulary with definitions All relations between the concepts o Hierarchical relations o Associative relations Difference compared to thesaurus, is the level of detail 11/53

12 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus o Ontologie Structuring methods o Taxonomy 12/53

13 Taxonomy 13/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Incident- beheer Configuratie- beheer TOPdesk 4 FAQ Aandachts- punten Incident- beheer Configuratie- beheer Incident- beheer Configuratie- beheer

14 Taxonomy disadvantages Knowledge item can not be used on multiple places in the structure 14/53

15 Taxonomy 15/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Incident- beheer Configuratie- beheer Kennis Item

16 Taxonomy disadvantages Knowledge item can not be used on multiple places in the structure Browsing can only be done mono- dimensional 16/53

17 Taxonomy 17/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Incident- beheer Configuratie- beheer Product: TOPdesk 4 Category: FAQ Productmodule: ?

18 Taxonomy 18/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Incident- beheer Configuratie- beheer Product: TOPdesk 4 Category: ? Productmodule: Incidentbeheer

19 Taxonomy 19/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Incident- beheer Configuratie- beheer Product: TOPdesk 4 Category: ? Productmodule: Incidentbeheer

20 Taxonomy disadvantages Knowledge item can not be used on multiple places in the structure Browsing can only be done mono- dimensional Knowledge items are a fundamental part of the structure 20/53

21 Taxonomy 21/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Incident- beheer Configuratie- beheer Archive FAQ

22 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus o Ontologie Structuring methods o Taxonomy o Tagging 22/53

23 Tagging Advantage: flexibility Disadvantage: will become a mess of definitions 23/53 Item FAQTOPdesk 4

24 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus o Ontologie Structuring methods o Taxonomy o Tagging o Wiki 24/53

25 Wiki Advantage: flexibility Disadvantage: gets a mess of relations and definitions 25/53 Item

26 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus o Ontologie Structuring methods o Taxonomy o Tagging o Wiki o Topic maps (Pepper, 2000) 26/53

27 Topic map 27/53 TOPdesk 4 FAQ Incident- beheer Configuratie- beheer Aandachts- punten Kennis Item Product Categorie Productonderdeel Product: TOPdesk 4 Category: ? Productmodule: Incidentbeheer Kennis Item

28 Awareness of problem Domain modelling schemes o Taxonomy o Thesaurus o Ontologie Structuring methods o Taxonomy o Tagging o Wiki o Topic maps (Pepper, 2000) o Knowledge entry maps (Helms; Brinkkemper; v Oosterum; Nijs; 2005) 28/53

29 Knowledge entry map 29/53

30 Agenda Background Awareness of problem Research question Research method Results Conclusions Discussion and future work 30/53

31 Research question ‘Is knowledge retrieval in a multipurpose knowledge repository using a specialized structuring method for a multipurpose knowledge repository, more efficient than knowledge retrieval in a multipurpose knowledge repository using a taxonomy structure?’ 31/53

32 Agenda Background Awareness of problem Research question Research method Results Conclusions Discussion and future work 32/53

33 Research method Design research 33/53 Takeda et al. (1990) Awareness of problem Suggestion

34 Existing structuring methods TOPdesk 34/53

35 TOPdesk TOPdesk company o Software developer from Delft o 250+ employees o Offices in Delft, Kaiserslautern, London TOPdesk product o Service management o Standard product o Important modules Incident management Configuration management Knowledge base 35/53

36 Suggestion Existing structuring methods TOPdesk Requirements engineering o 10 semi-structured interviews with users, developers, sales people, consultants o 2 Joint requirements sessions o Analysed incidents registered by the helpdesk o Analysis of the current system o Recorded in ‘natural language’ o Built an Acces application o Development of scenarios 36/53

37 Research method Design research 37/53 Takeda et al. (1990) Awareness of problem Suggestion Develop- ment Develop- ment Evaluation

38 Scenarios structured with structuring method in prototype Validation through test sessions o Stratified sample group of 20 persons 3 categories of web experience o 4 knowledge retrieval assignments o 2 systems Current TOPdesk knowledge system Taxonomy Prototype with new structuring method o Exactly the same test conditions o Measurement Amount of Time Amount of Clicks 38/53

39 Research method Design research 39/53 Takeda et al. (1990) Awareness of problem Suggestion Develop- ment Develop- ment Evaluation Conclusion

40 Agenda Background Awareness of problem Research question Research method Results Conclusions Discussion and future work 40/53

41 Results: Requirements 41/53

42 Results: Scenarios 2 scenarios “Small and simple knowledge domain” o 1 system administrator o Limited knowledge for setting up a structure o A few persons adding knowledge o Need for a flexible structure “Large complex knowledge domain” o Multiple system administrators o A lot of knowledge for setting up a structure o Hundreds of persons adding knowledge o Need for a compelling structure 42/53

43 Results: Development Solution: a “topic map” that can be created from within an application, which also allows a flexible structure, and relationships between knowledge items 43/53 CharacteristicTopic maps KEM WikiTagging Structure without set-upXX Knowledge item to knowledge item relationship X Topdown browsingX Compelling structureX

44 Results: Development Multipurpose Knowledge Repository Structuring Method (MKRSM) model 44/53

45 Results: Development 45/53

46 Results: Development Multipurpose Knowledge Repository Structuring Method (MKRSM) meta- model 46/53

47 Results: Prototype 47/53

48 Results: Validation 48/53 Amount of clicks Amount of time Scenario 1: Small and simple knowledge domain Results: Validation and Evaluation MAC Assignments 1 Taxono- my 1 MKRS- M 3 2 3,62,93,73 AVG 22,119,421,520,3 Average time between 2 clicks taxonomy structure: 6,0 s Average time between 2 clicks MKRSM structure: 6,7 s

49 Results: Validation and Evaluation 49/53 Amount of time Scenario 2: Large and complex knowledge domain Amount of clicks 9,154,5510,24,4 66,743,770,745 Average time between 2 clicks taxonomy structure: 6,9 s Average time between 2 clicks MKRSM structure: 8,8 s

50 Agenda Background Awareness of problem Research question Research method Results Conclusions Discussion and future work 50/53

51 Conclusions The developed structuring method improves knowledge retrieval efficiency in a multipurpose knowledge repository compared to a taxonomy structure. 51/53

52 Discussion and futurework Premise that sample group was a realistic representation of the population Testing maintainability Search engine remains important Automated suggestions for associations Technical design Standardized structures 52/53

53 Questions? »More information: »www.larsvandermeer.nl/thesis 53/53


Download ppt "Designing and validating a structuring method for a multipurpose knowledge repository And evaluating the method in a software prototype Lars van der Meer."

Similar presentations


Ads by Google