EMOTIONALLY INTELLIGENT KNOWLEDGE

EMOTIONALLY INTELLIGENT KNOWLEDGE SHARING BEHAVIOUR MODEL FOR CONSTRUCTING PSYCHOLOGICALLY AND EMOTIONALLY FIT RESEARCH TEAMS
R. KHOSLA AND M. HEDJVANI
Research Centre for Computers, Communication and Social Innovation School of Management, La Trobe University,
Melbourne, Vic 3086, Australia
E-mail: r.khosla@latrobe.eud.au
K.G. YAMADA, K. KUNEIDA AND S. OGA
C&C Innovation Res. Labs. NEC Corporation
8916-47 Takayama-cho, Ikoma-Shi, Nara, 630-0101, Japan.
E-mail: kg-yamada@cp.jp.nec.com
Knowledge sharing is an important driver for innovation in research teams and organizations. This paper views knowledge sharing as occurring in a quasi knowledge market of buyers and sellers. It makes unique contributions in terms of i) constructing a knowledge sharing behavior model based on different categories of knowledge sellers/buyers; ii) outlines application of non-invasive method for measuring the transient emotional state changes of a knowledge worker while they are being evaluated on their knowledge sharing behavior, iii) describes novel method for determining psychological and emotional fitness of knowledge workers to facilitate team or organization innovation and iv) design of emotionally intelligent knowledge management systems involving cognitive & non-verbal or emotional information.
Keywords: Knowledge sharing behavior model, Innovation, Knowledge market, Emotional states
1. Introduction
Innovation is the primary enabler for the organizations or teams to succeed in present competitive era (Davila, Epstein and Shelton 2006). Knowledge sharing is a vehicle for nurturing innovation and knowledge creation in teams and organizations (Nonaka & Takeuchi, 1995). In order to facilitate effective group innovation in organizations among knowledge workers it is important to analyze their knowledge sharing behavior and determine their psychological and emotional fitness in a team/organization. Researchers in the knowledge management community (Davenport & Prusak, 1998) have established that knowledge sharing occurs in a quasi knowledge market of buyers and sellers. This research models the knowledge sharing behavior of knowledge workers or researchers based on this well established perspective. A unique aspect of this research is that it considers psychological as well as emotional components of knowledge sharing.
Psychological and emotion profiling components of an Emotionally Intelligent Knowledge Sharing Behavior Modeling System (EIKSBMS) are described. A two dimension four category based knowledge sharing behavior model is developed. It is underpinned in constructs of a quasi knowledge market, namely, knowledge buying and selling. Fourteen areas are used to evaluate the knowledge sharing behavior of a knowledge worker or a researcher. The evaluation of knowledge workers using this model enables development of team or organization specific benchmarks for comparing knowledge sharing behavior profiles of two or more knowledge workers. The two dimensional, four category knowledge sharing behavior model based on existing work in the selling and buying behavior reported in (Davenport & Prusak, 1998; Buzzotte, Lefton & Sherberg, 1981). This model is in contrast to other studies which have examined the effect of personality traits on the knowledge sharing behavior (Hsu, Wu & Yeh, 2007; Matzler et al., 2008). This includes Big-5 personality model, the concept which was originally triggered by Thurstone (1934). Most of these studies have analyzed personality traits in a generic manner rather than in a direct manner in the context of quasi knowledge market (as reported in this paper). In the context of knowledge sharing behavior researchers have looked at use of Theory of Reasoned Action (TRA) Theory of Planned Behavior (TPB) for modeling knowledge sharing intent rather than developing distinct categories of knowledge sharing behavior types. Thus it is difficult to use their work for developing organization specific knowledge sharing behavior profile benchmarks for mixing and matching people in constructing psychologically compatible research teams. In the absence of behavior categories it may also be difficult to determine other psychological information like their motivation needs from a management perspective. Secondly, existing TRA and TPB related studies are not underpinned in a quasi knowledge market of knowledge buyers and sellers developed by Davenport and Prusak (1998), which is also the focus of this paper. Thirdly, existing TRA and TPB related studies have examined knowledge sharing intent in limited contexts and corpus of questions (e.g. 20 questions used by Ryu et al. (2003)). The limited corpus of questions can be ineffective because the human subjects or knowledge workers can get away with masked behavior in responding to the questions. One requires a much larger corpus of constructs and questions to get a reliable pattern of behavioral commitment of the knowledge workers. Additionally, studies with limited corpus or contexts do not facilitate adequate benchmarking or comparison of knowledge sharing behavior in different contexts. This is important in terms of identifying the training needs of knowledge workers in different contexts once they become part of a new team.
In terms of emotional component, non-verbal data forms an important component of human communication and behavior (Mehrabian, 1972; Picard, 1997). 55% of our communication is said to be through facial expressions and body gestures (Mehrabian, 1972). The second part of this paper involves modeling of non-verbal emotional responses of knowledge workers using a web camera while they are being evaluated on their knowledge sharing behavior. This enables comparison of emotional profiles of two or more knowledge workers for determining emotional fitness or cohesiveness in a research team based on their knowledge sharing behavior. It also enables correlation of emotional responses with cognitive responses provided by the knowledge worker while being evaluated on their knowledge sharing behavior using the four category behavioral model. Among other aspects, the correlation improves the information quality for distinguishing between knowledge workers in terms of their emotional drive and motivation. The paper is structured as follows: Section 2 discusses the theoretical underpinnings related to psychological component of knowledge sharing behavior. Section 3 outlines the psychology based knowledge sharing behavior model. Section 4 discusses the theoretical underpinnings related to emotion profiling component of knowledge sharing behavior. Section 5 briefly outlines the approach and methodology for design of emotionally intelligent knowledge sharing behavior model for constructing psychologically and emotionally fit research teams. Section 6 describes some interesting implementation aspects related to benchmarking, psychological and emotional fitness. Section 7 concludes the paper.
2. Knowledge Sharing Behavior Psychological Component - Theoretical Underpinnings
This section describes the relationship between knowledge sharing and innovation as the prime driver for constructing the knowledge sharing behavior model and the grounding of the knowledge sharing behavior model in quasi knowledge market of knowledge buyers and sellers.
2.1 Knowledge Sharing and Innovation

Figure 1 - The linkage betweens KS, Innovation and Value creation (Saenz, Aramburu & Rivera, 2009)
Innovation can be defined from variety of perspectives (Gopalakrishnan & Damanpour, 1997; Herkema, 2003; Abou-Zeid & Cheng, 2004; Plessis, 2007). This research investigates and models relationship between knowledge sharing (an element of knowledge management) and innovation (Darroch & McNaughton, 2002; Hong, Hwang & Lin, 2003; Saenz, Aramburu & Rivera, 2009). Since our study is on research teams, and the final product of research teams would be primarily innovation, we define innovation as the radical or incremental process of sharing and acquiring knowledge, transferring and transforming it to form new ideas or breakthroughs. Saenz et al (2009) stated that there is a link between knowledge sharing (KS) mechanisms and the firms’ innovation capability. They considered three main mechanisms for knowledge sharing which are “people-focused KS”, “IT based KS” and “management process-anchored KS” as shown in Figure 1. They divided the innovation capability constructs into three items including “new idea”, “innovation project management” and “time and cost efficiency”. The reason for this was to look further into the innovation implementation and the value which is resulted from that. Having a knowledge vision is valuable for supporting the “ideation” (Davila, Epstein and Shelton 2006) according to this study. In this work we aim to further examine the people focused knowledge sharing mechanisms and particularly psychological motivation theory.
2.2 Knowledge Market of Buyers and Sellers: a quasi market
Knowledge market is the place, in which knowledge is being exchanged, bought, found and generated. The main actors of knowledge market are buyers (knowledge seekers, users), sellers (knowledge providers), brokers and even sometimes entrepreneurs who take the advantage of the knowledge as the source of power and value across the firm. In knowledge market, the pricing system is to an extent determined by participants gaining the utility and value for themselves and their team or organization (Davenport & Prusak, 1998; Cross & Prusak, 2003). According to Davenport and Prusak (1998), the knowledge market in organization is highly dynamic and is known to be a kind of “quasi market”, a market in which exchange of goods can’t be forced by agreements and formal contracts. Although in knowledge market we observe both the actors buyer and seller, there is a fundamental difference compared to the conventional market. In knowledge market the buyer is more eager to buy the product (i.e. knowledge) whereas in conventional market seller is generally more willing to sell their product. Another difference is changing nature of the role of sellers and buyers in the knowledge market through the time. Considering the research team, one person can be the knowledge seller at one time and knowledge buyer at another time.
3. Knowledge Sharing Behavior Model
In this research the focus is how to construct psychologically fit and emotionally cohesive research teams based on people focused knowledge sharing behavior model as against IT focused agent mediated knowledge market modeling (Zmud, 1984). The knowledge sharing behavior model shown in Figure 2 is conceptually underpinned in two concepts, a quasi market of knowledge buyers and sellers discussed in the last section and psychological factors or personal drives or need/s which motivate different types of knowledge sharing behaviors in a quasi market of knowledge.
The concept of knowledge buyers and sellers is analogous to interaction based selling and buying of a product with the difference that the product here is knowledge. So in this research the authors have built upon the work in behavioral psychology in studying need driven psychological factors which govern selling and buying behavior (Buzzotte, Lefton & Sherberg, 1981; Khosla, Goonesekera & Chu, 2008) and developed a model for knowledge sharing behavior. The knowledge sharing behavior model is shown in Figure 2 and consists of two dimensions and four behavior categories. These two dimensions are “warm-hostile,” and submissive-dominant.” The two dimensions "Submissive------Dominant" and "Warm-----Hostile" are the two most significant dimensions in which knowledge selling and buying behavior is expressed. These two dimensions give rise to four broad groups of knowledge workers, i.e., Dominant-Hostile (DH), Submissive-Hostile (SH), Submissive-Warm (SW), and Dominant-Warm (DW).The description for each category for the knowledge sellers is summarized in Figure 2.


























Figure 2: Knowledge Sharing Behavior Model
Warmth is regard for others. A warm person is optimistic and willing to place confidence in others. Hostility is lack of regard for others, the attitude that other people matter less than oneself. A hostile person rarely trusts others. Submission is the disposition to let others take the lead in personal encounters. It includes traits like dependence, unassertiveness, and passiveness. Dominance is the drive to take control in face-to-face situations. It includes a cluster of traits like initiative, forcefulness, and independence. The reasons for using this particular model are a) the domain experts found it less complex, b) they found it easy to relate with as it mimicked their way thinking for typifying/categorizing inter-personal behavior among knowledge workers and, c) they found this model close to inter-personal behavior training programs they had undergone.
One could use more dimensions like IQ, temperament, etc which could be associated with the model. However, it was felt a) it would make the model more complex, b) marginalize distinctions between knowledge workers and make documenting the knowledge more difficult, and c) it would be better develop a subsystem, with only these additional dimensions and put the conclusions of the two systems together.
The behavioral descriptions of the four categories are shown in Figure 2. These descriptions are shown from knowledge buyer as well as knowledge seller perspective. The behavioral model and four categories are also related to Abraham Maslow's model of hierarchy of human (personal) needs which determine the motivating needs or personal drives of knowledge workers. For example, a SH knowledge worker is driven by security and biological needs which form the lowest level of unfilled need. A SW knowledge worker is driven more by social needs and less by security needs. They believe all knowledge workers are well meaning and their needs to socialize and befriend people prevent them from discriminating among knowledge workers on the basis knowledge value or benefit. On the other hand, a DH knowledge worker is driven by independence and control needs. The common hostile dimension in both SH and DH leads them not to trust other fellow knowledge workers. However, both adopt different strategies to deal with their lack of trust of others. A SH knowledge worker avoids getting involved in knowledge sharing with other, whereas, DH knowledge worker’s control needs drive them to demonstrate their superiority over other knowledge workers. Finally, as per the model the DW knowledge worker represents the ideal knowledge sharing behavior, wherein they engage in knowledge sharing with other knowledge workers based on knowledge value and mutual benefit. The pursuit of knowledge value is their mean of satisfying their high level need for self-realization.
The behavioral categories and their needs can also be understood and correlated in the context of the pricing system elements of the knowledge market defined by Davenport and Prusak (1998) and psychology driven motive system defined by McClelland (1985). The pricing system for knowledge market described by emphasized that in knowledge market participants are concerned about the expected utility rather than just money. There are four constructs introduced as the possible forces which shape the knowledge market pricing system: reciprocity, reputation, altruism and trust. The psychology driven motive system identifies the possible sources of motivation in human being. These sources or categories are “achievement, power, and affiliation (i.e. love) and avoidance (i.e. fear) related motives (McClelland, 1985). There is a clear match between this categorization and the knowledge market pricing system constructs. Table 1 shows the correlation between four behavior categories, the four pricing system elements of knowledge market as well as the four psychological motive groups based on the motive system theory. A SH knowledge worker’s behavior is underpinned in lack of trust of other workers resulting in avoidance and lack of involvement in knowledge sharing. On the other hand, a SW person is driven by need to socialization and intimacy which are reflected in altruism and affiliation. They believe that altruism can’t exist without love among people and the need for trust in the environment can clearly solve problems which are connected to avoidance related issues. The DH knowledge worker is motivated by needs of independence and control which are satisfied through power and reputation as a proof of superiority of knowledge. Finally, DW knowledge worker is motivated by need for self-realization which is reflected through sense of achievement and reciprocity in terms of mutual benefit of knowledge. DW category is mostly searching for gaining tangible benefits from knowledge sharing .Reciprocity can be seen as a form of achievement related motive for enhancing mutual benefit and knowledge value
4. Knowledge Sharing Emotion Profiling Component - Theoretical Underpinnings
It is useful to understand the possible link between cognitive responses of a user and their emotions from the perspective human communication, human behavior and correlation between cognitive and emotional responses. Several models of human emotions have been devised by researchers (Sloman, 1987; Ortony, Clore & Collins, 1994; Izard, 1990; Roseman, Antoniou & Jose, 1996). The model which compares favorably with findings in psychological, cognitive science and neurobiological communities is the Sloman’s three layer information processing architecture (Picard, 1997).
Behavior Category of Knowledge Sharing Model Knowledge Market Pricing Element (Davenport and Prusak [8] ) Motive System Category
(McClelland [23] )
Dominant-Hostile (DH) (independence and control) Reputation Power
Dominant Warm (DW) (self-realization) Reciprocity Achievement
Submissive-Hostile (SH) (security) (lack of) Trust Avoidance
Submissive Warm (SW) (socialization) Altruism Affiliation (love)
Table 1: Correspondence between Knowledge sharing
Behavior category, knowledge market pricing system and
motive system
In Sloman’s architecture, as the first layer, the reactive Layer detects thing in its environment, and executes fairly automatic processes to determine how to react. The deliberative Layer is capable of planning, evaluating options, making decisions, and allocating resources. The emotions involved in goal-success or goal-failure, i.e., those which are cognitively assessed, are also found in this layer. The third layer, Self-Monitoring Meta-Management Layer, prevents certain goal from interfering with each other, and can look for more efficient ways for the deliberative layer to operate, choose strategies and allocate its resources. In particular, it illustrates the need for a higher “self-monitoring” process for management of emotions. The latter is a crucial piece of a system if it is to develop the skills of emotional intelligence for regulating and wisely using its emotions (Picard, 1997).
The EIKSBMS described in this paper evaluates a knowledge worker in terms of their notion of self in context of their knowledge sharing behavior. That is, it evaluates their disposition, attitude or beliefs towards fourteen different areas related to research, peers, etc. There are 60 selling behavioral questions in all and at least 4 questions in each area (related to 4 selling behavior categories). Some areas have two sets of 4 questions. The questions in each area are deliberately designed to contradict each other in order to facilitate a pattern of commitment in the responses. Based on the nature of the application and evaluation, the emotions are triggered by the self-monitoring layer (as it is challenged by the contradictory nature of attitude/belief based questions) and expressed through physiological indicators like facial expressions. An affect space model developed by psychologists (Lang, 1995) and employed for facial expression analysis (Picard, 1997; Cohn & Kanade, 2006) is shown in Table 2. The model involves three dimensions, namely, Valance (measured on a scale of pleasure (+) to displeasure (-)), Arousal (measured on scale of excited/aroused (+) to sleepy (-)) and Stance (measured on a scale of high confidence (+) and low confidence (-)). Facial expressions correspond to affect states like happy, surprise and tired. Figure 3 shows the affect space model with several labeled emotional states. Like in everyday life, in human-computer interaction people’s emotions are characterized more by subtle variations or transient changes in facial/emotional expressions (during the interaction) rather than as prototypical emotional expressions (Edwardson, 2000). It can be noted from Table 2 that emotional state like anger, sadness and fear are not being specifically measured. In other words, subtle variations or changes are modeled using positive, negative and neutral (no change) states. The positive state is represented by positive emotional state quadrant of the affect face model shown in Table 2. The negative state is represented by negative emotional state quadrant in Table 2. The neutral state shown in Table 2 represents the area which is enclosed by the original face model of a human subject. The model can thus be divided into quadrants, each quadrant being considered to represent positive or negative emotional states. Note that what we are modeling here is change in emotional state with time and whether this change is in a direction towards a positive or negative quadrant of the affect space model.
5. Approach and Methodology
This research has been conducted along three dimensions, namely, field studies, generalizability and modeling precision. The field studies involve analysis of actual knowledge sharing behavior and emotional profile of knowledge workers in research institutions and ICT industry using the knowledge sharing behavior model ( Figure 2) and affect space model (Table 2) and facial action coding system respectively. The generalizability dimension involves a random survey of knowledge workers in ICT industry. The primary purpose of the random survey is to establish the reliability of the 60 measured items (i.e., knowledge sharing behavior questions in the survey). The field studies and random survey are in process at the time of writing this paper.
The third dimension involves development of knowledge sharing behavior model shown in Figure 2 and emotional state and intensity profile of knowledge workers (based on affect space model in figure 3 and Facial Action coding systems (Ekman & Friesen, 1978)) for measurement of 14 independent variables associated with knowledge sharing behavior and 4 dependent variables related to knowledge behavior categories, and 36 independent variables associated with facial features (or facial action units) and 4 dependent variables ( emotional state changes (+ive, -ive, neutral) and intensity). The rest of this section outlines briefly the methodology for the design and analysis of these independent and dependent variables
5.1 Knowledge Sharing Behavior Psychological Component Analysis Phase
Very few knowledge workers in the real world could be expected to be perfect fits in any category. In fact the behavioral profile of most of them will have parts in each category. Thus although the behavioral model can provide us some basis for distinction, it cannot be used as a conclusive proof of a knowledge workers primary selling knowledge sharing behavior It is here that the role of the domain experts becomes extremely important. How do they use this knowledge in a manner which helps them to deal effectively with knowledge workers? What areas they feel are important for gauging knowledge sharing behavior? What areas are overlooked or considered unimportant? Answers to these questions would provide us the basis for evaluating and determining a knowledge worker’s primary or predominant knowledge sharing behavior category. The primary or predominant behavior category is the category which determines a knowledge worker’s interactions with other knowledge worker’s in knowledge selling and buying scenarios. It further establishes their corresponding unfilled personal or motivating need related to the predominant behavior category.
















The areas for evaluation of a knowledge workers’ knowledge sharing behavior are shown in Figure 3
Questions Behavior Category
1. You treat knowledge sharing with every peer as a challenge which must be met successfully.
DH
2. You determine who are best peers for knowledge sharing and devote your time and energy on them only.
DW
3. You do your job. You do not significantly interact with your peers SH
4. You believe knowledge sharing follows from friendship. Since one person’s vote is as good as another’s, you do not discriminate among peers SW

Figure 3: 14 Areas for Gauging Knowledge Sharing Behavior and Sample Questions Related to Attitude Towards Peers or Team Members
These areas have been identified after several discussions with research team leaders and areas like trust and job satisfaction identified by other researchers (Ryu, Ho & Han, 2003; Kuoa & Young, 2008). After determining the different areas and their weights, attributes related to each of these areas with respect to different behavioral categories have been determined. The attributes of each of these areas have been designed in the form of questions. At least four questions have been designed for each area (one belonging to each behavior category) based on the knowledge sharing behavior model. The other parameters that have been kept in view while designing the questions are tone, length, total number, ordering and the pattern of questions. A sample set of four questions related to the area of competition is shown in Figure 4 and each question is related to one of the four behavioral categories. In order to quantify the varying degree of importance attached to the different areas of selling behavior by the domain experts (research managers), weights have been assigned to them on a scale of 1 to 10 using AHP technique. In order to determine the primary behavioral category of the knowledge worker the accumulated answer score in each behavioral category on all the questions is calculated using the formula:


i=14
∑ = [Area Weight _i]*[Answering Option Percentage Weight]
i=1
5.2 Knowledge Sharing Behavior Emotion Component Analysis Phase
A standard webcam has been used to assist in pupil tracking and normalization of facial images. Gabor wavelets has been used among several facial expression modeling techniques for tracking the changes in the facial action units (Akamatsu et al., 1998; Calder et al., 2001; Lee, 1996). Fuzzy rules have been used for inferencing neutral, +ive and –ive states and emotional state intensity based on facial action parameters involving angle and magnitude of movement of facial features like eye brows, cheeks and lips.

6. Implementation – Benchmarking, Psychological and Emotional fitness
This section is divided into two parts, namely, Benchmarking and Psychological Fitness, and emotional fitness and correlation of psychological and emotional profiles.
6.1 Benchmarking and Psychological Fitness
Figure 4(i) shows a sample comparison of pruned scores of two subjects in four behaviour categories (SH, SW, DH and DW). The comparison can be used for organisation or team specific benchmarking (or determining psychological fitness of a new knowledge workers’ in context of knowledge sharing behaviour profile against an existing knowledge worker (or research team member shown as benchmark profile in Figure 4(i) who represents the desired knowledge sharing behavioural attributes). The comparison indicates a high degree of correlation or fitness between the benchmark profile and new knowledge worker (candidate profile in Figure 4(i)). In other words, the candidate will have less training needs if included in the team and represents a high level of psychological fitness in terms of knowledge sharing behaviour and consequently can contribute effectively towards group or organisational innovation.
6.2. Emotional Fitness and Correlation of Emotional and Psychological Profiles
In order to determine emotional fitness and correlation between cognitive and emotional responses in context of knowledge sharing behaviour, a high resolution web based camera is used to capture the video sequence of a candidate answering the questions.
The expressions presented are not contrived, that is, the expressions are genuine responses to the questions being presented. Figure 4 (ii) also shows the difference images and visualizations of the neural network classification. The sequence shown is taken at a time between a new question being presented to the candidate and the candidate answering that question. Gabor wavelet and neural network classifier are used for processing and classifying the negative, positive and neutral emotional state of the sales candidate (Abou-Zeid & Cheng, 2004). The expression in Figure 14 (ii) (a) was classified as a +ive emotional response with low emotional intensity as it represents a mix of roughly equal proportions of neutral and positive as indicated by the blue and green respectively and absence of red (which represents negative emotion change). The expression in Figure 14 (ii) (b) was classified as primarily neutral indicated by the diminished green in (b) and the shift from cyan to a more blue colour in the top half of the classification image with respect to (a). The expression in Figure 14 (ii) (c) classification indicates a dominance of red and is classifies as –ive emotional state change with high emotional intensity because of absence of any other colour representing +ive or neutral states. These +ive, -ive emotional state changes are used to construct emotional profile of the candidate for all the questions. The emotional profile is then compared with emotional profiles of other team members in order to determine whether the new team member or candidate lies within highest and lowest emotional band based on emotional profiles of other team members.










(i)

(iii)
(ii)
Figure 14: (i) Comparison of Candidate’s profile with Benchmark profile based on cognitive responses (answers to questions), (ii) Classification of +ive, neutral and –ive Emotional State Changes Based on Sequence of Images from Video. (iii) Correlation of Emotional (ES A I – Emotional State And Intensity) with Cognitive Responses (CR) in Context of Attitude Towards Peers and Manager (team leader) respectively.
Experiments both in the field and in laboratory have confirmed that either input (i.e., cognitive/keyboard or emotional) provides incomplete and inaccurate information for knowledge computation and interpretation. The correlation of the emotional state profile and knowledge sharing behavior profile is done at 3 levels. Level 1 correlation involves correlating the emotional state profile with knowledge sharing behavior profile of the candidate in each area of knowledge sharing behavior evaluation (shown in Figure 3). As mentioned in section 2 each area the candidate is evaluated based on 4 questions related to DH, SW, SH, and DW knowledge sharing behavior categories respectively. The correlation at level 1, among other aspects, provides insight into which knowledge sharing behavior dimension is more prevalent in the candidate in one or more areas. For example, if emotional states positive (with high intensity) and positive (med) are related to affirmative answers to DH and SH questions in Figure 3 then they indicate that the Hostile (H) dimension is more emphasized in the candidate than the Dominant (D) dimension in the area of attitude towards peers. Level 2 correlations involve correlating the emotional state and behavior profiles based on each behavioral category. Level 3, firstly, involves fine grain correlation related to all the 60 questions. Secondly and finally, it involves emotional profile and selling behavior profile correlation and comparison with organization specific or organization defined benchmark/s.
The level 1 correlation is shown in Figure 14 (iii). They are used to determine good and less good correlations between cognitive and emotional responses in 14 areas of evaluation. The numbers (0, 1, 2 and 3) on X-axis represent the four questions (corresponding to four behavioral categories, DW, DH, SH, SW) in each area (e.g. attitude towards manager). The numbers (1, 0, -1) on Y-axis represent ‘Yes’, ‘Not Sure’ and ‘No’ answer or Cognitive Response (CR) to a question. Cognitive responses “To a Large Extent Yes’ and ‘To a Large Extent No’ are represented in between 0 and 1, and 0 and -1 respectively. The numbers 1, 0 and -1 also represent +ive, neutral and –ive Emotional State (ES) responses are corresponding to a cognitive response. The ranges 0 to 1 and 0 to -1 also captures the intensity of the +ive and –ive emotional state responses respectively. For example, in Figure 14 (iii) the point (-1, 2) in the ‘Attitude towards Manager’ graph represents a good correlation corresponding to the question “You welcome criticism from your boss…” That is, it shows to a negative cognitive response and a negative emotional state response. On the other hand, a positive cognitive response (point (1, 0.5) in the ‘Attitude towards Peers’ graph) and a negative emotional response (point (1,-1) the ‘Attitude towards Peers’ graph) indicate a less good correlation. The comparison between two candidates can be done based on number of similar or dissimilar good and less good correlations and also in terms of similar or dissimilar emotional state intensity. The good and less good correlations can also be used to customize the interview of a candidate. For example, good correlation tends to confirm that a candidate firmly believes in the scenario portrayed by a question corresponding to a particular area and behavioral category. A less good correlation means the candidate may be probed on the particular question during the interview.

7. Conclusion
Knowledge sharing is a vehicle for nurturing innovation and knowledge creation in teams and organizations. Knowledge sharing occurs between knowledge buyers and knowledge sellers in a knowledge market. Measurement of knowledge sharing behavior involves both rational and affective or emotional characteristics of knowledge workers. The rational characteristics are influenced by psychologically driven personal needs and subjective norms or culture of an organization. This paper describes a unique and novel method for modeling psychological and emotional fitness of a knowledge worker in a research team environment. These two parameters can be used to construct research teams to enhance team and organizational innovation and developing team or organization specific psychological and emotional fitness benchmarks. In the process it develops a two dimension four category knowledge sharing behavior model of knowledge buyers and sellers. This paper also heralds a new way of designing emotionally intelligent knowledge management systems which involve analysis and correlation of cognitive and non-verbal or emotional data where ever people are involved.

References

Abou-Zeid, E., & Cheng, Q. (2004). The effectiveness of innovation: a knowledge management approach, International Journal of Innovation Management, 8(3), pp. 261–274
Akamatsu, S., Gyoba, J., Kamachi, M., & Lyons, M. (1998). Coding facial expressions with Gabor wavelets”, Automatic Face and Gesture Recognition, Proceedings of Third IEEE International Conference, pp: 200 – 205, 4-16 April 1998.
Buzzotte, V. R., Lefton, R. E., & Sherberg, M. (1981). Effective Selling Through Psychology, Psychological Associates New York
Calder, A. J., Burton, A. M., Miller, P., Young, A. W., &Akamatsu, S. (2001). A principal component analysis of facial expressions. Vision Research, 41, pp. 1179-1208
Cohn, J. and Kanade, T. (2006). Use of Automated Facial Image Analysis for Measurement of Emotion Expression, The Handbook of Emotion Elicitation and Assessment. Oxford University Press Series in Affective Science, J. A. Coan & J. B. Allen, ed., 2006
Cross, R., & Prusak, L. (2003) The political economy of knowledge market, in Easterby-Smith, Mark and Marjorie A. Lyles, eds. Handbook of Organizational Learning and Knowledge Management. Oxford: Blackwell Publishing, pp. 454-72
Darroch, J, R, McNaughton (2002) Examining the link between knowledge management practices and types of innovation, Journal of Intellectual Capital; 3(3) ABI/INFORM Global, pg. 210
Davenport, T.H., Prusak, L. (1998). Working knowledge: How organizations manage what they know, Harvard Business School Press, Boston, Massachusetts, USA
Davila, T., Epstein, MJ., & Shelton, R (2006). Making Innovation Work: How to Manage it, Measure it, and Profit from it, Pearson Education, Upper Saddle River, NJ., USA
Edwardson, M. (2000). Emotion Profiles of customers and their impact on marketing strategy, Customer Loyalty, May 2000, Sydney.
Ekman, P., & Friesen, W. (1978). Facial Action Cording System: A Technique for the Measurement of Facial Movement, Consulting Psychologists Press, Palo Alto, CA
Gopalakrishnan, S., & Damanpour, F. (1997). A review of innovation research in economics, sociology and technology management, Omega: The International Journal of Management Science, 25(1), 15–28
Herkema, S. (2003) A complex adaptive perspective on learning within innovation projects, The Learning Organization, 10(6), pp. 340-346.
Hong, JC., Hwang, MY., & CL, Lin. (2003). Chi and Organizational Creativity: A Case Study of Three Taiwanese Computer Firms, Creativity and innovation management, 10(4), pp. 202-210
Hsu, BF., Wu, WL., & Yeh, RS. (2007). Personality Composition, Affective Tie and Knowledge Sharing: A Team Level Analysis, PICMET 2007 Proceedings, 5-9 August, Portland, Oregon – USA
Izard, C. E. (1990). Facial Expressions and the Regulation of Emotions, Journal of Personality and Social Psychology, 58 (3), pp. 487-498
Khosla, R., Goonesekera, T., & Chu, M. (2008). Separating the wheat from the chaff: An intelligent sales recruitment and benchmarking system, Expert Systems with Applications, 36, pp.3017–3027
Kuoa, F.Y., & Young, M.L. (2008). Predicting knowledge sharing practices through intention: A test of competing models, Computers in Human Behavior, 24(6), pp. 2697-2722
Lang, P.J. (1995). The Emotion Probe: Studies of Motivation and Attention, American Psychologist, 50(5), pp. 372-385
Lee, T. (1996). Image Representation Using 2D Gabor Wavelets, IEEE Transactions on Pattern Analysis and Machine Intelligence, 18(10):9590971, Oct. 1996.
Matzler, K., Renzl, B., Müller, J., Herting, S., & Mooradian, T.A. (2008). Personality traits and knowledge sharing, Journal of Economic Psychology, 29, pp. 301–313
McClelland, DC. (1985). Human Motivation, Scott: Freshman and Company, USA
Mehrabian, A. (1972). Nonverbal communication, Aldine Transactions, USA
Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company, Oxford University Press, Oxford
Ortony, A., Clore, G. L., & Collins, A. (1994) The Cognitive Structure of Emotions, Cambridge University Press.
Picard, R.A., (1997). Affective computing, MIT Press
Plessis, M D. (2007). The role of knowledge management in Innovation, Journal of knowledge management, 11(4), pp. 20-29
Roseman, I. J., Antoniou, A. A., & Jose, P. E. (1996). Appraisal determinants of emotions: Constructing a more accurate and comprehensive theory. Cognition and Emotion, 10, 241-277.
Ryu, S., Ho, S. H., & Han, I. (2003). Knowledge sharing behavior of physicians in hospitals, Expert Systems with Applications, 25 (1), pp. 113-122.
Saenz, J., Aramburu, N., & Rivera, O. (2009). Knowledge sharing and innovation performance: A comparison between high-tech and low-tech Companies, Journal of Intellectual Capital, 10(1), pp. 22-36
Sloman, A. (1987). Motives, Mechanisms, and Emotions. Cognition and Emotion, 1, 3: 217-33.
Thurstone, LL. (1934). The vector of the mind, Psychological Review, 41(1)
Zmud, RW. (1984). An Examination of 'Push-Pull' Theory Applied to Process Innovation in Knowledge Work, Management Science, 30 (6), pp. 727-738