Analyze cognitive process of information requirement analysis.

Huang, I.-Lin

INTRODUCTION

Information requirement analysis is the early phase of information systems development. During information requirement analysis, information analysts capture, understand, and translate users' information requirements into requirement specifications (Gibson & Conheeney, 1995; Huang, 2008). The resulting requirement specifications have at least three purposes: (1) facilitating an understanding of the intended system, (2) guiding the process of information system design, and (3) serving as a basis for all communications concerning the information system being developed (Hsia, Davis, & Kung, 1993; Schemer, 1987).

The correctness of requirement specifications is important for the success of an information system development project. An estimation showed that inaccurate requirement specifications might cost in excess of one hundred times what would have been required if the errors were discovered during information requirement analysis (Roman, April 1985; Shemer, 1987). A similar survey done by the Standish Group (1995) also showed that 31.1% of software projects in the United States were cancelled at some point during the development cycle; and inaccurate or incomplete requirement specifications were identified as the most important contributing cause. Therefore, how to specify correct requirement specifications is a critical issue for information requirement analysis.

Information requirement analysis is an error prone process, especially for novice information analysts. Empirical studies have shown that lack of knowledge is a major cause for novice information analysts making more errors in requirement specifications (Schenk, Vitalari, & Davis, 1998). Empirical studies have also shown that four characteristics of modeling behaviors that set expert and novice information analysts apart: model-based reasoning, mental simulation, critical testing of hypotheses, and analogical domain knowledge reuse (Sutcliffe & Maiden, 1990). However, it is unclear how the knowledge of information analysts may influence their modeling behaviors in information requirement analysis. Therefore, the research question of this research is "What is the cognitive process model of information requirement analysis that can explain how the differences of knowledge of information analysts may lead to different modeling behaviors?"

In this article, a cognitive process model of information requirement analysis is constructed on the basis of the structure-mapping model of analogy. On the basis of the cognitive process model of information requirement analysis, the interactions between the knowledge of information analysts and modeling behaviors are explained from the perspective of the dynamic process of information requirement analysis.

The remainder of this paper is organized as follows. First, this research will review the empirical studies related to the knowledge and modeling behaviors of information analysts. Then this research will discuss Gentner's structure-mapping model of analogy and explain why it is a good choice as a basis for modeling the cognitive process of information requirement analysis. Third, on the basis of the structure-building model of analogy, this research will propose a cognitive process model of information requirement analysis. Fourth, this research will use the proposed cognitive process model to explicate the differences between novice and expert information analysts in information requirement analysis. Fifth, this research will discuss the implications of the cognitive process model for research and practices in information requirement analysis. Finally, a conclusion will be made in the final section.

LITERATURE REVIEW

This section will first review the research studies concerning the influence of the knowledge of information analysts on the performance of information requirement analysis. Then, the review will discuss the literature on the differences of modeling behaviors between expert and novice information analysts. On the basis of the findings, we will explore the important cognitive processes of information requirement analysis in the following sections.

THE KNOWLEDGE OF INFORMATION ANALYSTS

The research into the influence of the knowledge of information analysts on the performance of information requirement analysis has been conducted in two categories: knowledge availability and knowledge organization (Schenk, Vitalari, & Davis, 1998). Knowledge availability refers to various types of knowledge used in information requirement analysis. On the other hand, knowledge organization refers to the ways by which the knowledge is stored in the long-term memory of information analysts.

Knowledge Availability

Domain knowledge and modeling knowledge have been suggested as determining factors for the modeling performance of information analysts. Domain knowledge is drawn upon by both expert and novice information analysts in specifying information requirements (Sutcliffe & Maiden, 1990; Vessey & Conger, 1993). While understanding problem statements, information analysts use domain knowledge to mentally simulate a scenario of the system behavior in order to test the adequacy of the requirement specifications, to add assumptions to increase the completeness of the requirements, to test internal and external consistency of the requirements, and to abstract, summarize, select and highlight important information in the problem statements (Guindon, Krasnar, & Curtis, 1987). Without domain knowledge, even expert information analysts can only specify high-level conceptual models without details (Adelson & Soloway, 1985). With the availability of domain knowledge, novice information analysts can reuse the domain knowledge to achieve almost the same level of completeness of requirement specifications as expert information analysts do (Sutcliffe & Maiden, 1990).

On the other hand, modeling knowledge has long been regarded as an important factor to differentiate expert from novice information analysts. Modeling knowledge can be divided into syntactic and semantic parts (Koubek, et al., 1989). Syntactic knowledge consists of allowable syntax of a specific modeling language. Semantic knowledge, however, consists of modeling principles that are independent of a particular modeling language (Allwood, 1986). Compared to novice information analysts, expert information analysts with richer semantic knowledge can retrieve and apply more relevant modeling principles, make more critical testing of hypotheses, and finally achieve requirement specifications with better quality (Allwood, 1986; Koubek, et al., 1989; Schenk, Vitalari, & Davis, 1998; Vitalari & Dickson, 1983). Modeling knowledge can also be divided into declarative and procedural aspects (Vessey & Conger, 1993). The procedural aspect of a requirement analysis technique is more difficult to learn than the declarative aspect. However, the procedural aspect of modeling knowledge is more important in determining the quality of requirement specifications (Vessey & Coger, 1993).

Knowledge Organization.

There are basically two features of knowledge organization that can differentiate expert from novice information analysts in information requirement analysis: the size of knowledge unit and the level of abstraction of knowledge. First, expert information analysts store their knowledge in bigger units than novice information analysts do. Empirical studies showed that storing knowledge in bigger chunks gives expert information analysts advantages over novice information analysts in understanding and specifying information requirements. First of all, experts can automate some aspects of the problem solving process because their knowledge can be mapped onto a problem context in a bigger scope. As a result, expert information analysts can have a more efficient process of information requirement analysis. On the other hand, novices have to solve the problem from the first principle due to smaller units of knowledge in the memory. Novice information analysts have to spend much more cognitive resources in identifying the relevant pieces of knowledge and put them together in the right way, leading to an inefficient process of information requirement analysis (Allwood, 1986; Guindon, Krasner, & Curtis, 1987; Guinder & Curtis, 1988). Even worse, smaller units of knowledge may make the process of problem solving more complicated for novice information analysts. As a result, many errors can be caused by novices' inability to map parts of the problem description to appropriate knowledge structures as well as by novices' failure to integrate pieces of information (Allwood, 1986).

The second feature is that expert information analysts use higher-order abstract constructs to organize large amounts of knowledge while novice analysts store concrete objects sparsely in the long-term memory. Research evidence shows that experts use richer vocabulary to categorize problem descriptions into standard abstraction. As a result, experts can retrieve knowledge structure easily, and they can focus more on the semantic structure of problems rather than the surface or syntactic structure (Allwood, 1986; Koubek, Salvendy, Dunsmore, & Lebold, 1989).

Due to the above two important features of knowledge organization, expert analysts can have better performance in information requirement analysis by (1) processing large amounts of information into meaningful chunks; (2) retrieving the knowledge structure easily; and (3) categorizing problems into standard types based on underlying domain principles (Batra & Davis; 1992).

THE MODELING BEHAVIORS OF INFORMATION ANALYSTS

Empirical research on the cognitive process of information requirement analysis has identified a strong association among the activities of gathering information, identifying relevant facts, and conceptual modeling (Batra & Davis, 1992; Sutcliffe & Maiden, 1992). This strong association reflects that information requirement analysis is basically an understanding process.

To account for the better performance of expert information analysts in understanding and specifying information requirements, the research on cognitive process has focused on the differences in the modeling behaviors between expert and novice information analysts. Empirical studies on the modeling behaviors of information analysts showed that four modeling behaviors set expert and novice information analysts apart: model-based reasoning, mental simulation, critical testing of hypotheses, and analogical domain knowledge reuse.

First, expert information analysts use model-based reasoning to model information requirements with the help of various requirement analysis techniques (Sutcliffe & Maiden, 1992; Vitalari & Dickson, 1983). Research evidence showed that model-based reasoning on the basis of requirement analysis techniques could produce more complete solutions than partial or no model-based reasoning behavior. On the other hand, research evidence also showed that novice information analysts could not perform model-based reasoning effectively because they had difficulties in identifying important concepts in the requirement statements by requirement analysis techniques (Sutcliffe & Maiden, 1992). For example, in a research study on the modeling behaviors of novice information analysts in using data flow diagrams, it was shown that the novice information analysts were more successful at recognizing system goals and inputs, while there was poorer recognition of system data stores, processes, and outputs, even though data stores, processes, and outputs were explicitly stated in the problem narrative (Sutcliffe & Maiden, 1992). Therefore, we may conclude that effective model-based reasoning is an important cognitive process that sets expert and novice information analysts apart.

The second feature of expert analysts' modeling behaviors is mental simulation. Mental simulation refers to the cognitive processes of building a mental model that establishes connections among the parts of the system under investigation and of using the mental model to reason about the interactions among the parts of the system (Adelson & Soloway, 1985; Guindon, Krasner, & Curtis, 1987; Guinder & Curtis, 1988). During information requirement analysis, expert information analysts use requirement analysis techniques for mental simulation of information requirements while novice analysts used requirement analysis techniques only for representation (Adelson & Soloway, 1985). Mental simulation makes expert analysts focus on the semantic part of the problem statement. On the other hand, without mental simulation novice information analysts can analyze only the syntactic part of the representation (Adelson & Soloway, 1985; Allwood, 1986).

Critical testing of hypotheses is the third feature of the modeling behaviors of expert information analysts. By means of mental simulation, expert information analysts can have a clear picture about the structure of the information requirements (Guindon, Krasner, & Curtis, 1987; Guinder & Curtis, 1988). Consequently, experts may be more able to reason about a problem, to create test cases and scenarios for testing hypotheses critically (Schenk, Vitalari, & Davis, 1998; Vitalari & Dickson, 1983). On the other hand, novice information analysts can generate hypotheses only at a general level and make few attempts to test hypotheses because they focus only on the syntactic part of the representation (Schenk, Vitalari, & Davis, 1998).

Finally, analogical domain knowledge reuse makes expert information analysts able to specify information requirements more completely and accurately (Mainden & Sutcliffe, 1992). Expert information analysts tend to use higher-order abstract constructs to organize large amounts of knowledge. As a result, expert information analysts can recognize and assimilate analogies more easily (Batra & Davis, 1992; Vitalari & Dickson, 1983). In addition, expert information analysts tend to keep in memory the details of requirement specifications from their past experience. Consequently, higher quality can be expected because the reused specifications are well tested and validated. On the other hand, novice information analysts have difficulty in identifying the opportunities of analogical modeling because they tend to store concrete objects sparsely in the long-term memory (Batra & Davis, 1992; Sutcliffe & Maiden, 1992). In addition, novice information analysts tend to specify information requirements from scratch because of the lack of reusable specifications in their memory (Vitalari & Dickson, 1983).

THE STRUCTURE-MAPPING MODEL OF ANALOGY

Gentner's structure-mapping model of analogy will be used in this research as the basis for the cognitive process model of information requirement analysis because of the following two reasons: First, the output of the structure-mapping model of analogy is a situation model of the problem context under investigation, which is the same as the output by the cognitive processes of text comprehension and information requirement analysis. Due to the common cognitive goals, the structure-mapping model may be able to shed more light on the cognitive process of information requirement analysis from the perspective of text comprehension. Second, the strength of the structure-mapping model is its ability to explain the differences of analogical reasoning between novices and experts (Gentner, 1983). According to the structure- mapping model, experts use structural similarity as the basis for analogical reasoning and hence can get better understanding of the target phenomenon. On the other hand, novices use attribute similarity as the basis for analogical reasoning and hence cannot get correct interpretation of the target phenomenon (Gentner, 1983). Therefore, the structure-mapping model may be able to explicate the issue of novice-expert differences better in information requirement analysis. In this section I will discuss the structure-mapping model of analogy from the perspectives of the following four characteristics: (1) the task, (2) the assumption, (3) the mapping process, and (4) the guiding principle for mapping process.

The Task

There are two domains, target domain and base domain, in the context of analogy. The task of analogy is to define a mapping from B, which is a concept in the base domain, to T, which is a concept in the target domain. When the mapping is done, we can conclude the analogy by saying that "T is (like) B". In this analogy, T will be called the target because it is the concept that we want to comprehend. B will be called the base because it is the concept that we know very well and hence that serves as a source of knowledge.

The Assumption

In order to explain the cognitive process of analogy by the structure-mapping model, Gentner (1983, pp. 156-157) made four assumptions about the cognitive environment: (1) "Domain and situations are psychologically viewed as systems of objects, object-attributes, and relations between objects." On the basis of this assumption, Gentner limited the elements of a conceptual structure to three constructs: object, attribute, and relation. (2) "Knowledge is represented as propositional networks of nodes and predicates." This assumption limited the knowledge organization in memory as propositions (Kintsch & Dijk, 1978), rather than schema (Schank & Abelson, 1977) or Neuro-network (Kintsch, 1988). (3) "Two essentially syntactic distinctions among predicate types will be important. The first distinction is between object attribute and relationships. Attributes are predicates taking one argument, and relations are predicates taking two or more arguments. The second important syntactic distinction is between first order predicates (taking objects as arguments) and second- and higher-order predicates (taking propositions as arguments)." The purpose of this assumption is to design a computing mechanism for explicating the process of analogy reasoning. And finally (4) "These representations, including the distinctions between different kinds of predicates, are intended to reflect the way people construe a situation, rather than what is logically possible." This assumption express the concern of the structure-mapping model is the cognitive process of building a situation model, which is the same as that of text comprehension and information requirement analysis.

The Mapping Process

There are four kinds of domain comparison processes that can determine the mapping from a concept in the base domain to a concept in the target domain: literal similarity, analogy, abstraction, and surface similarity. First, literal similarity is a comparison in which a base structure can be mapped onto the target structure with both object-attributes and structural (or called relational) predicates. For example, Monkey feet are like human feet. In this comparison, monkey's feet are not only similar to human feet in attributes (toe, shape, etc.) but also in structural predicate (for walking, jumping, and supporting body).

Second, analogy is a comparison in which structural predicates, but few or no object attributes, can be mapped from base to target. For example, Cars are like human feet. In this example, cars are different from feet in attributes; but similar in structural predicates (for transportation, for example).

Third, abstraction is a comparison in which the base structure is an abstraction of the target structure. For example, cars are transportation devices.

Fourth and finally, surface similarity is a comparison in which base structure share similar objects and attributes with the target knowledge. For example, we may say that cars are like bricks because of similar shape.

The Guiding Principle for the Mapping Process

While mapping the base structure onto the target structure, a higher-order relation (or predicate) will be more likely to be imported into the target structure than is an isolated relation or object-attribute. It is called the principle of systematicity (Gentner, 1983). This principle is derived from the fact that human beings pursue coherent situation model during their comprehension process. A higher-order relation defines a structure connecting more concepts and lower-level relations together than an isolated relation or object-attribute does. As a result, a higher-order relation contributes higher coherence to the situation model of the problem context and hence provides more satisfaction for the comprehenders.

A COGNITIVE PROCESS MODEL OF INFORMATION REQUIREMENT ANALYSIS

On the basis of the structure-mapping model of analogy (Falkenhainer, Forbus, & Gentner, 1990; Gentner, 1983; Gentner & Markman, 1997), this research proposes a cognitive process model of information requirement analysis to explicate the modeling behaviors of information analysts as shown in Figure 1 (Huang & Burns, 2000). In this section, this research will discuss the mechanism of this cognitive process model. The strength of this model that can explain the interactions between the knowledge of information analysts and different modeling behaviors between novice and expert information analysts will be discussed in the next section.

In this section, we will assume a requirement sentence, "The customer first sends an order to John, the order clerk," in a problem statement of an order processing system as an example to illustrate the cognitive process of information requirement modeling. On the basis of the cognitive process model depicted in Figure 1, the cognitive process of information requirement analysis can be divided into three parts: parsing, modeling, and questioning as follows.

PARSING

A problem statement is the source of target structures that includes concepts and structures of information requirements. The task of information requirement analysis is to construct a model that can connect all concepts and structures of the problem statement into a coherent whole. If a coherent model can be built for the problem statement, then the task of understanding the problem statement is achieved.

Parsing as the first step in modeling translates the example sentence into a target structure in the form of propositional knowledge as follows (Kintsch, 1974):

send (CUSTOMER, ORDER, ORDER CLERK)

send : predicate; CUSTOMER: agent; ORDER: object; and ORDER CLERK: agent.

[FIGURE 1 OMITTED]

The translation depends on analysts' knowledge mainly about natural language (in this case, English) and domain knowledge. In this article, we assume that both novice and expert information analysts have the same level of ability to understand English text and necessary domain knowledge about an ordering system. Thus, we can assume that both novice and expert information analysts can come up with a piece of propositional knowledge similar to the above one.

MODELING

Modeling is the process that translates the received target structure into the form of a base structure of a particular requirement analysis technique. In this article, we assume that the selected requirement analysis technique is the data flow diagrams. On the basis of the cognitive process model, the modeling process can be divided into three subprocesses: accessing, mapping and evaluating as follows:

Base Structure Selection

In order to specify the information requirements in the problem statement by a particular requirement analysis technique, information analysts access the base structures of the requirement analysis technique to match the incoming target structure. Basically, two factors are considered while determining which base structure will be selected: the principle of continuity (Zwaan, Graesser, & Magliano, 1995) and the types of similarity (Gentner, 1983). First, on the basis of the principle of continuity, information analysts tend to access the base structure that can be connected to the submodels that have been built so far, especially the most recent one. This principle reflects that information analysts try to build a connected and coherent model for the whole problem statement.

Second, there are four types of similarity between target and base structures that can trigger the access of a particular base structure: literal similarity, analogy, abstraction, and surface similarity. First, on the basis of literal similarity, the information analyst may decide that the order processing system under investigation is like that of company X I analyzed last year. Second, on the basis of analogy, the information analyst may conclude that the order processing system is like the library system he or she analyzed two years ago. Third, abstraction reasoning may make the information analyst use the base structure, inflow (external entity, dataflow, process), from data flow diagrams to model the target structure. Fourth and finally, surface similarity may attract the information analyst's attention and decide too model customer as external entity, and order as data store.

Empirical evidence shows that human knowledge is more likely organized by object-attribute similarity, rather than by structural similarity. Thus, novice information analysts tend to access base structures by literal similarity or surface similarity because both have the feature of object-attribute similarity. Abstraction and analogy are rarely used by novice information analysts to access base structures because the structural similarity is more difficult to identify.

On the other hand, expert information analysts have learned from experience that structural similarity (or even higher-order structure similarity) has better explanation power than object-attribute similarity. Therefore, expert information analysts will prefer abstraction and analogy to surface similarity in selecting base structures. Empirical evidence shows that experts learn from experience to organize their knowledge by abstract relations rather than objects or attributes (Halford, 1987).

For illustration, if the information analysts decide to use the data flow diagrams to model the example sentence mentioned above, the expert information analysts may select a higher-order relational base structure like inflow (external entity, data flow, process). On the other hand, novice information analysts may select an object-attribute base structure like external entity, data store, and external entity to match the three concepts in the problem statement: CUSTOMER, ORDER, and ORDER CLERK.

Structure Mapping

While mapping the base structure onto the target structure, a higher-order relation (or predicate) will be more likely to be imported into the target structure than is an isolated relation or object-attribute on the basis of the principle of systematicity. For example, if the selected based structure is inflow (external entity, data flow, process), then the information analyst will be able to get the following three results on the basis of model-based reasoning:

CUSTOMER will be mapped as external entity, and ORDER as data flow;

ORDER CLERK cannot be mapped as process. The information analyst may therefore make inferences to decide that the process is what the order clerk does--order processing; and the information analyst may find out by abstraction that the requirement "customer first sends an order to the order clerk" is an input data flow for a high-order structure--an order processing system. On the basis of the principle of systematicity, the information analyst may try to model the whole order processing system by identifying data stores and output data flows from his or her domain knowledge.

Coherence Evaluation

The result submodel will finally be evaluated on the basis of coherence. For example, by using the base structure inflow (external entity, date flow, process) to match the requirement sentence send (CUSTOMER, ORDER, ORDER CLERK), we will find ORDER CLERK can not be matched by process because ORDER CLERK is obviously an agent rather than a process. If the information analyst cannot identify "processing order" as the process by model-based reasoning, then the mismatch between ORDER CLERK and "process" will cause an incoherence. Consequently, the information analyst may decide to abandon the mapping and try another base structure; or he may choose to keep it and solve the incoherence later.

QUESTIONS GENERATING: ASKING QUESTIONS ABOUT THE INCOHERENCES IN THE SUBMODEL

The incoherences in submodels will become the cues for questioning (Huang, 2006). For example, in order to erase the incoherence on the mismatch between ORDER Clerk and "process," information analysts may ask questions to identify the missing process in the submodel. Example question may be like:

What task is done by the order clerk? Or more directly, what is the process for the incoming order?

AN EXPLANATION FOR THE NOVICE-EXPERT DIFFERENCES

The purpose of the proposed cognitive process model of information requirement analysis is to describe the modeling behaviors of information analysts. The cognitive process model argues that the differences of knowledge availability and knowledge organization determine the different modeling behaviors between expert and novice information analysts. The different modeling behaviors, in turn, lead to the different levels of correctness of requirement specifications.

The strength of the cognitive process model is its ability to explain an unclear issue related to the performance of information requirement analysis: how the differences of knowledge between novice and expert analysts may lead to different modeling behaviors from the perspectives of four characteristics: model-based reasoning, mental simulation, critical testing of hypotheses, and analogical domain knowledge reuse? The explanation based on the cognitive process model is as below:

First, how does the knowledge of information analysts influence the model-based reasoning? The purpose of model-based reasoning is to identify the concepts for requirement specifications correctly and completely (Sutcliffe & Maiden, 1992). Expert information analysts organize their knowledge by abstract relations. Thus, expert information analysts can make model-based reasoning effectively because they access base structures for modeling target structures on the basis of structural similarity. Consequently, expert information analysts can get fewer errors in their requirement specifications. On the other hand, novice information analysts organize their knowledge as concrete objects sparsely in the long-term memory. Thus, they select base structures on the basis of object-attribute similarity that will be more likely to cause errors or incomplete concepts in the requirement specifications (Sutcliffe & Maiden, 1992).

Second, how does the knowledge of information analysts influence mental simulation? The purpose of mental simulation is to reason about the interactions among the parts of a system and then to establish coherent connections among the parts for a more complete requirement specification (Adelson $ Soloway, 1985; Guindon, Krasner & Curtis, 1987; Guindon & Curtis, 1988). Expert information analysts organize their base structures in bigger units that have higher coherence. The higher coherence will, in turn, provide richer explanation power for mental simulation while modeling the target structures. As a result, fewer errors will be generated in their requirement specifications. On the other hand, novice information analysts have their base structures in smaller units that will result in many small fragments of requirement specifications. Even worse, many of the smaller requirement fragments may be generated on the basis of object- attribute similarity. As a result, the limited or even wrong explanation power will make the mental simulation difficult and thus many errors will be generated during the integration of requirement fragments into bigger and more complete requirement specifications.

Third, how does the knowledge of information analysts influence critical testing of hypotheses? Critical testing of hypotheses is important to validate the coherence of requirement specifications. On the basis of base structures with higher abstraction and bigger unit, expert information analysts can make critical testing of hypotheses more effectively to derive more important concepts on the basis of the principle of continuity. As a result, more complete requirement specifications can be generated. On the other hand, with a model built from object- attribute similarity, novice information analysts can generate hypotheses only at a general level and make few attempts to test hypotheses (Sutcliffe & Maiden, 1992).

Fourth and finally, how does the knowledge of information analysts influence the performance of analogical domain knowledge reuse? On the basis of the principle of systematicity, expert information analysts can identify opportunities of analogical reasoning more easily because they use abstract concepts to organize their knowledge. In addition, expert information analysts can reuse specifications in bigger units and with higher quality because they store in memory the details of the well tested and validated specifications from their past analysis experience. On the contrary, novice information analysts have difficulty in identifying analogies because they focus on concrete objects and attributes. As a result, they often need to develop requirement specifications on the basis of the first principle.

IMPLICATIONS OF THE COGNITIVE PROCESS MODEL

The cognitive process model has shown that knowledge of information analysts lead to different modeling behaviors and different modeling behaviors in turn result in differences in the correctness of requirement specifications. The cognitive process model suggests that the most basic reason accounting for the differences between novice and expert information analysts is that novice and expert information analysts pay attention to different aspects of a problem statement: experts focus on the structural side of the problem statement but novices on the object-attribute side. Therefore, at least two implications can be identified in this research: first, in order to accelerate the transition from novice to expert information analysts, novice information analysts should be encouraged to learn and to think in terms of structures rather than of object-attributes. Actually, thinking in terms of structures has also been suggested as an effective way to improve students' reading comprehension (Nix, 1985). Second, novice information analysts can have the same level of performance as expert information analysts have if the target and the base structures share literal similarity that includes both structural and object-attribute similarities. Therefore, domain-specific requirement analysis techniques deserve future research because they use the same concepts and structures as those of the problem statements and hence will improve the productivity of novice information analysts significantly.

CONCLUSION

Using structure-building model of analogy as a reference, this research has proposed a cognitive process model of information requirement analysis. The structure-building model of analogy as a reference model has provided the proposed cognitive process model with two advantages. First, the cognitive model focuses on the process of building situation model for the problem context under investigation, which is consistent with the concern of the cognitive process of information requirement analysis. Second, the cognitive model focuses on the differences between expert and novice in modeling behaviors, which is also the major concern of the research in cognitive process of information requirement analysis.

The cognitive process model proposed in this research has explicated the interactions among the cognitive variables from the perspective of dynamic process of information requirement analysis. In addition, by linking the knowledge of information analysts with the modeling behaviors of information analysts, the cognitive model provides the theoretical explanation about why novice and expert information analysts have different modeling behaviors during information requirement analysis. Finally, the cognitive process model has also shown that the structural similarity between users' problem statements and requirement analysis techniques is an important determinant for the degree of difficulty in information requirement modeling.

REFERENCES

Adelson, B., & Soloway, E. (1985). The role of domain experience in software design. IEEE Transactions On Software Engineering, 11 (11), 1351-1360.

Allwood, C. M. (1986). Novices on the computer: a review of the literature. International Journal of Man-Machine Studies, 25, 633-658.

Batra, D., & Davis, J. G. (1992). Conceptual data modeling in database design: similarities and differences between expert and novice designers. International Journal of Man-Machine Studies, 37, 83-101.

Falkenhainer, B., Forbus, K. D., & Gentner, D. (1990). The structure mapping engine: algorithm and examples. Artificial Intelligence, 41, 1-63.

Gentner, D. (1983). Structure-mapping: a theoretical framework for analogy. Cognitive Science, 7, 155-170.

Gentner, D., & Markman, A. B. (1997). Structure mapping in analogy and similarity. American Psychologist, 52 (1), 45-56.

Gibson, M.D., & Conheeney, K. (1995). Domain knowledge reuse during requirements engineering. In Proceedings of CAiSE 1995 (pp. 283-296). Berlin: Springer-Verlag.

Guindon, R., Krasner, H., & Curtis, B. (1987). Cognitive process in software design: activities in early, upstream design. In H. J. Bullinger & B. Shackel (Eds), Human-computer Interaction, INTERACT'97 (pp. 383-388). Amsterdam: North-Holland.

Guindon, R., & Curtis, B. (1988). Control of cognitive process during software design: what tools are needed? In Proceedings of CHI'88 (pp. 263-268). New York: ACM Press.

Halford, G. S. (1987). A structure-mapping approach to cognitive development. International Journal of Psychology, 22, 609-642.

Hsia, P., Davis, A. M., & Kung, D. C. (1993, November). Status report: requirements engineering. IEEE Software, 75-79.

Huang, I (2006). A design of self-questioning mechanism for information requirement specification. Academy of Information and Management Sciences Journal, 9(1), 103-122.

Huang, I (2008). A cognitive explanation of the correctness of information requirement specifications. Academy of Information and Management Sciences Journal, 11(2), 1-18.

Huang, I., and Burns, J.R. (2000). A cognitive comparison of modeling behaviors between novice and expert information analysts. Proceedings of AIS Conference, 2000. 1316-1322.

Kintsch, W. (1974). The representation of meaning in memory. Hillsdale, NJ: Lawrence Erlbaum.

Kintsch, W. (1988). The role of knowledge in discourse comprehension: a construction-integration model. Psychological Review, 95 (2), 163-182.

Kintsch, W., & van Dijk, T. (1978). Toward a model of text comprehension and production. Psychological Review, 85 (5), 363-394.

Koubek, R. J., Salvendy, G., Dunsmor, H. E., & Lebold, W. K. (1989). Cognitive issues in the process of software development: review and reappraisal. International Journal of Man-Machine Studies, 30, 171-191.

Nix, D. H. (1985). Notes on the Efficacy of Questioning. In A. C. Graesser & J. B. Black (Eds.), The Psychology of Questions (pp. 297-333). Hillsdale, NJ: Lawrence Erlbaum.

Roman, G. (1985, April). A taxonomy of current issues in requirements engineering. IEEE Computer, 14-22.

Schank, R. C., & Abelson, R. P. (1977). Scripts, Plans, Goals and Understanding: an Inquiry into Human Knowledge Structures. Hillsdale, NJ: Lawrence Erlbaum.

Schenk, K. D., Vitalari, N. P., & Davis, K. S. (1998). Differences between novice and expert systems analysts: what do we know and what do we do? Journal of Management Information Systems, 15 (1), 9-50.

Shemer, I. (1987). System analysis: a systematic analysis of a conceptual model. Communications of the ACM, 30 (6), 506-512.

Standish Group. The Standish Group Report: CHAOS, Retrieved July 12, 2000, from http://www.scs.carleton.ca/~beau/PM/Standish-Report.html

Sutcliffe, A., & Maiden, N. (1990). Cognitive studies in software engineering. In Proceedings of the Fifth Conference on Cognitive Ergonomics (pp. 159-171). European Association of Cognitive Ergonomics.

Sutcliffe, A., & Maiden, N. (1992). Analysing the novice analyst: cognitive model in software engineering. International Journal of Man-Machine Studies, 36, 719-740.

Vessey, I., & Conger, S. A. (1993). Learning to specify information requirements: the relationship between application and methodology. Journal of Management Information Systems, 10 (2), 177-201.

Vitalari, N. P., & Dickson, G. W. (1983). Problem solving for effective system analysis: an experimental exploration. Communications of the ACM, 26 (11), 948-956.

Zwaan, R. A., Graesser, A. C., & Magliano, J. P. (1995). Dimensions of situation model construction in narrative comprehension. Journal of Experimental Psychology, 21 (2), 386-397.

I-Lin Huang, Langston University