ExpertLink: An Environment for Tacit Knowledge Management



ExpertLink: An Environment for Tacit Knowledge Management

Abstract

An effective way to manage tacit knowledge is to know “who knows what.” Usually an organization constructs a yellow page of expertise to provide such information. However, given the trend of organization globalization, the manual effort required for the creation of yellow page may prohibit the sharing of tacit knowledge. In addition, an individual may lack incentive to share because of the extra burden this may cause or the concern of losing power within the organization. On the other hand, archives of computer-mediated communication (CMC) not only document knowledge shared among participants but also record their behaviors. An active person during CMC may not be the most knowledgeable person but most likely he/she is willing to help. Knowing where and who those people are can be crucial to tacit knowledge sharing. This paper thus proposes the ExpertLink system that utilizes information representation, analysis, and visualization technologies to facilitate locating active persons. The proposed system develops a searching component that supports concept-based search for experts. Unlike conventional search engines that return related textual documents based on relevance, the ExpertLink system ranks search results based on both the relevance and the willingness to help. Although the system development is still at its early stage, we believe that the automatic approach proposed in this paper can support tacit knowledge sharing within an organization.

Key Words: Knowledge Management, Information Analysis, Information Visualization, Information retrieval, Computer-based Communication Systems.

1. Introduction

Between the two knowledge management strategies identified by Hansen, et al. (1999), the personalization strategy manages tacit knowledge residing in minds of individuals. Unlike explicit knowledge, which can be expressed or stored in natural language or metaphors, tacit knowledge is difficult for individuals to articulate (Nonoka & Konno, 1998). The personalization strategy emphasizes on building personal relationships such that individual expertise can be provided to the right people at the right time. In addition to connecting people through various computer-mediated communication tools (CMC), organizations usually provide yellow pages of expertise to help people locate the right expert. However, the creation of such yellow pages may involve manual efforts, which can be a barrier to tacit knowledge management due to the lack of incentives. There are usually two reasons for this. First, sharing knowledge with others may add extra workload. Secondly, an individual may perceive a loss of power by giving out what he/she knows. Therefore some organizations such as Ernst & Young reward employers for sharing. At the same time, more and more informal groups, called ‘community of practice’ (Wenger, 1998) have been formed to encourage the learning process among group members.

The crucial role of the computer-mediated communication in knowledge management stems not only from the fact that inter-personal communication enabled by a CMC system results in knowledge sharing and transfer (Sachs, 1995), but also from the increasing importance of CMC in maintaining a virtual community that facilitates the organizational learning process (Sproull & Kiesler, 1991). While a CMC process documents knowledge shared, it also records the behavior of its participants and their attitude toward the virtual community (Sproull & Kiesler, 1991; Weisband et al., 1995). We believe that the archives of CMC process are valuable resources for locating tacit knowledge. Participants who always provide answers to others in those virtual communities might not be the most knowledge, but they are probably willing to help. Knowing who and where those people are and what their specialties are can be helpful in tacit knowledge sharing. In addition, the CMC has been embedded into normal organizational process, applying information technologies to locate active persons from CMC archives does not add any extra work.

This paper thus proposes the ExpertLink system, an information system that recommends relevant persons from archives of CMC discussions based on both the message content and participants’ behaviors during discussions. The system is an extension of the CommunicationGarden system (Zhu & Chen, 2001), which combines information analysis technology with social visualization approach to provide a graphical representation of expertise. In addition to utilizing one component from the CommunicationGarden system, the ExpertLink system develops a searching component to help users with specific information needs. The searching component supports concept-based searching over archives of CMC and returns a list of people ranked according to the relevance to users’ search queries and the willingness to help. Technologies used by the ExpertLink system include nature language processing based noun phrasing (Tolle & Chen, 2000), co-occurrence analysis (Chen & Lynch, 1992), self-organizing map (SOM) (Kohonen, 1995) and social visualization (Donath, et al., 1999). The ExpertLink system will provide following functionalities:

1. The system automatically generates a graphical yellow page of expertise to facilitate the browsing behavior

2. The system supports concept-based search for people by understanding the meaning of query terms specified by its users.

2. Research Formulation

2.1. Computer Mediated Communication and Expertise Identification

Most empirical studies of CMC have focused on the interaction among information technology, individual behavior, group characteristics, and organizational structure. These studies indicate that CMC not only provides incentives for participants to share knowledge (Sproull & Kiesler, 1991) but also creates a common context in which its participants convert their tacit knowledge into explicit knowledge (Nonaka & Konno, 1998). At the same time, CMC participants also project their personal styles, previous experiences and social norms into their computer mediated communication (Weisband et al., 1995). The attitude of participants toward the community is related to the volume of messages they send (Sproull & Kiesler, 1991). In addition, the person who posts more answers or participates more in discussion of certain topics than do other individuals may be regarded as the expert in that area (Ahuja & Carley, 1998). He/she might not be the most knowledgeable individual on that subject but is probably willing to help. Knowing who and where those experts are is another type of valuable organizational knowledge. The archive of a CMC process therefore contains rich information about both knowledge shared and behavior of participants, which can be helpful to users wishing to locate people for direct contact.

2.2. Existing Systems

Although an archive of a CMC does contain information about the location of tacit knowledge, no system has been found to deliver such information to users. On one hand, there are systems organizing discussion content by mediating the way its participants communicate (Nunamaker et al, 1991; Ackerman, 1998) or by applying different information analysis and artificial intelligence technologies to facilitate browsing and searching behaviors over an archive (Konstan et al., 1997; Chen et al. 1998b; Van Dyke at al., 1999). On the other hand, systems such as Loom (Donath et al., 1999) and PeopleGarden (Xiong & Donath, 1999) use social visualization technologies to provide graphical summaries on such participants’ behaviors as who starts a discussion, who talks with whom, how long a person stays, and how lively a discussion is. Although the CommunicationGarden system provides summaries of both content and behavior, it still does not support the searching behavior of users with specific information need.

2.3.Vocabulary Problem

In addition to using yellow page of expertise, a user may also want find an expert through searching. The difference between searching and browsing is that a user usually has a specific goal in mind for searching. One issue confronted by the searching approach is the vocabulary problem. Individuals from different domains may use the same term for different meanings and use different terms for the same meaning. Because a user cannot select a domain, a search query may bring back information both within and outside the desired domain. For instance, for the search the term “cell,” a system may return information in both the domains of biology and mobile phone. The proposed system thus utilizes the co-occurrence algorithm to understand the semantics of query terms and to locate active person from the domain desired.

3. System Architecture and Technology Selection

The system architecture displayed in figure 1 was designed to support users’ browsing and searching behaviors. The system consists of four components: representation, backend databases, browsing component, and search component. Following is the detailed description of the four components.

3.1 Information Representation

Automatic indexing denotes to representing a document with a vector of terms automatically (Salton, 1989). Because the natural language processing (NLP) noun phrasing technique has been used in information retrieval to capture a richer linguistic representation of document content (Anick and Vaithyanathan 1997), the CommunicationGarden system selected one of the available NLP noun phrase tools, the Arizona Noun Phraser (AZNP) to represent the content of messages. The AZNP has been found to have better performance than other NLP noun phrase tools in identifying key noun phrases (Tolle & Chen, 2000). While a yellow page of expertise is created based on individuals’ behavior and message, the system regards messages sent by one person as a unit and represents each unit with key phrases identified by the AZNP.

3.2 Backend Databases

There are three types of database at the backend of this system: documents database, key term database, and people database. The information stored in the documents database includes the content of a message and its sender, date, length, etc. The key terms extracted from messages are stored in the key term database. The people database records information of CMC participants including name, e-mail, number of messages send out, number of discussion or thread participated, and time duration of staying in a CMC community.

3.3 Browsing Component

The browsing component integrates an information categorization technology, self-organizing map (SOM) with a social visualization technology to present a graphical yellow page of expertise. SOM is defined as a mapping from a high-dimensional input space into a two-dimensional array of output nodes, where spatial proximity represents semantic proximity. Several recent studies adopted the SOM approach to textual analysis. For instance, SOM has been used by Chen et al. (1998) to categorize and identify sub-topics from messages generated by electronic meeting systems. In this system, SOM was used only as a categorization tool and its spatial representation was not used.

Social visualization research represents human behavior graphically. Systems such as Loom (Donath et al., 1999) and PeopleGarden (Xiong & Donath, 1999), for instance, provide graphical summaries on who starts a discussion, who talks with whom, how long a person stays, and how lively a discussion is. The ExpertLink system utilizes the floral representation developed in Zhu & Chen (2001) to depict how a CMC participant behaves during the communication process. Figure 2 displays an example of such yellow page of expertise, where the entire interface is divided into sub-gardens based on subject categories identified by the SOM. Within a sub-garden, one flower represents a person, while the number of petals equals the number of messages that the person has posted. The number of leaves indicates the number of threads in which the person has participated, and the height of the flower represents how long the person stayed in the community. The graphical interface suggests not only how active each category is but also who is the most active person in each category. A user may locate a person of interest by browsing the graphical interface and selecting blooming flowers in a topic of interest. The effectiveness of this interface in delivering information about active person in certain topic has been demonstrated in Zhu & Chen (2001).

3.3 Searching Component

When a user has specific goal in the mind, he/she might want specify his/her information needs direct to the system. Two algorithms have been used in the searching component: the co-occurrence analysis and the results ranking.

• Co-occurrence analysis creates a concept space by identifying relationships among terms. The created concept space can help a user to refine a query by providing a set of related terms to the keyword provided by the user. More detailed description about this algorithm is provided in Chen & Lynch (1992). This technique has been applied in different domains, including Russian computing (Chen & Lynch, 1992) and group support systems (Chen et al., 1993). ExpertLink will utilize the concept space created to return related terms from different domains. According to the terms selected by the user, the system can locate the domain desired by the user thus search for active people from that domain only.

• After the system identifies a list of persons who have posted messages related to the users’ information need, the system will rank the list based on the both the relevance and the willingness to help. How related a person is to users’ search query will be measured by the number of occurrence of search terms and number of related messages this person has contributed. The willingness to help can be measured by the number of answers this person has provided to other participants. At this time, each factor has the same weight in the ranking algorithm. An evaluation study will be conducted to decide the weights of each factor.

4. Summaries and Discussions

The archives of CMC not only record explicit knowledge shared among participants, but also contain behavioral information about participants’. Those two types of information may suggest the locations of tacit knowledge. The ExpertLink system utilizes information representation, analysis, and visualization to help users locate appropriate person to have a direct contact. Adopted from the CommunicationGarden system, the browsing component of the ExpertScopre system provides a graphical yellow page of expertise to facilitate browsing behavior. The searching component, on the other hand, supports the concept-based searching for experts. A ranking algorithm is also developed to rank returned results based both the relevance and the willingness to help. Although the ExpertLink system is still at its early stage, we believe that the proposed system may have positive impact on managing tacit knowledge. While an efficient way of tacit knowledge management is to know “who knows what,” the proposed system helps users identify individuals who may not be the most knowledgable but are helpful. A manager can also use this system to recommend experts to employees. In addition, this system may also help to address the incentive issue confronted by tacit knowledge management. The CMC has been embedded into normal working process and it records all the information automatically, the automatic approach proposed thus can provide information about tacit knowledge without adding extra burden.

The technologies utilized by the ExpertLink system have also been shown feasible by previous studies. For instance, the combination of SOM and floral representation was shown to be effective in presenting who is the most active person in certain topic by Zhu & Chen (2001). In addition, the concept space generated by using co-occurrence analysis has also been used by various studies to facilitate searching behaviors (Zhu et al., 1999; Chau et. al., 2002). Each technical component is available to us and we are in the process of integrating them using the logical architecture shown in figure 3. Several empirical studies will be conducted after the implementation of the ExpertLink system including the validation the ranking algorithms [pic]

developed, evaluation of the usability of ExpertLink system, and the identification of users’ strategy of using the system.

References

1. Ackerman M. S. (1994). “Augmenting the organizational memory: A field study of answer garden,” Proceedings of CSCW’94, pp. 243-252.

2. Ahuja, M. K. and Carley, K. M. (1998). “Network structure in virtual organizations,” Journal of Computer-Mediated Communication, Vol. 3,

3. Anick, P. G. & Vaithyanathan (1997), Exploiting clustering and phrases for context-based information retrieval, In the 20th Annual International ACM SIGIR Conference on Research and Development, Philadelphia, PA, pp.314-323.

4. Chau, M., Chen, H., Qin, J., Zhou, Y., Qin, Y., Sung, W., and McDonald, D (2002). “Novel search environments: Comparison of two approaches to building a vertical search tool,” Proceeding of the second ACM/IEEE-CS joint conference on Digital libraries.

5. Chen, H. & Lynch, K.J., (1992). Automatic construction of networks of concepts characterizing document database, IEEE Transactions on Systems, Man and Cybernetics, Vol. 22, No. 5, pp. 885-902.

6. Chen, H., Titkova, O., Orwig, R., and Nunamaker, J. F. (1998), “Information Visualization for Collaborative Computing,” IEEE Computer, Vol. 31, No. 8, pp. 75-82, August, 1998.

7. MIS Quarterly, Volume 12, Number 2, 1988, pp. 259-275.

8. Donath, J., Karahalios, K., and Viegas, F. (1999), “Visualizing conversation,” Journal of Computer-Mediated Communication, Vol. 4,

9. Hansen, M. T., Nohria, N., & Tierney, T. (1999), What’s your strategy for managing knowledge, Harvard Business Review, March-April, pp. 106-116.

10. Konstan, J. A., Miller, B. N., Maltz, D., Herlocker, J. L., Gordon, L. R., and Riedl, J. T. (1997). “GroupLens: Applying collaborative filtering to Usenet News,” Communication of the ACM, Vol. 40, no. 3, pp. 77-87.

11. Kohonen, T. (1995). Self-organized Maps, chapter 3. Springer-Verlag, Berlin Heidelberg.

12. Nonaka, I. and Konno, N. (1998). “The concept of ‘Ba’: Building a foundation for knowledge creation,” California Management Review. Vol. 40, no. 3, pp. 40-55.

13. Nunamaker, J. F., Dennis, A. R., Valacich, J. S., Vogel, D. R., and George, J. F. (1991). “Electronic meeting systems to support group work,” Communications of the ACM. Vol. 34, no. 7, July, pp. 40-61.

14. Sachs, P. (1995). “Transforming work: collaboration, learning, and design,” Communications of ACM, Vol. 38, No. 5, pp. 36-45

15. Salton, G. (1989), Automatic Text Processing, Addison-Wesley Publishing Company, Inc., Reading, MA.

16. Sproull, L. and Kiesler, S. (1991). “Computers, Network, and Work,” Scientific American, Vol. 265, no. 3, September 1991, pp. 116-127.

17. Subramani, M. & Hahn, J. (2000). “Examining the effectiveness of electronic group communication technologies: The role of the conversation interface,” presented at Academy of Management Conference.

18. Tolle, K. M. and Chen, H. (2000). “Comparing noun phrasing techniques for use with medical digital library tools,” Journal of the American Society for Information Science, Vol. 51, No. 4, pp. 352-370.

19. Van Dyke, N. W., Lieberman, H., and Maes, P. (1999), Butterfly: A conversation-finding agent for internet relay chat, Proceedings of the 1999 International Conference on Intelligent User Interfaces, Redondo Beach: CA, pp. 629-644

20. Weisband, S. P., Schneider, S. K., and Connolly, T. (1995), Computer-mediated communication and social information: Status salience and status differences, Academy of Management Journal, Vol. 38, No. 4, pp. 1124-1151.

21. Xiong, R. & Donath J. (1999), Creating data portraits for users, Proceedings of the 12th annual ACM symposium on User Interface Software and Technology, pp. 37-44.

22. Zhu, B., Ramsey, M., Chen, H., Hauck, R.V., Ng, T. D., and Schatz, B. (1999). "Support Concept-Based Multimedia Information Retrieval," Proceedings of ICIS'99, 20th Annual International Conference on Information System.

23. Zhu, B. and Chen, H. “Social Visualization for Computer-Mediated Communication: A Knowledge Management Perspective,” accepted by the Eleventh Workshop On Information Technologies And Systems (WITS'01).

-----------------------

[pic]

Figure 1 System Architecture

[pic]

Figure 2 Interface of the Browsing Component

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download