(pressing HOME will start a new search)

Integration of a Database Management System and a Knowledge-Based Expert System in Construction: A Review

Chao Chen, Philip D. Udo-Inyang, and Frederick C. Schmitt

Department of Civil Engineering

Temple University

Philadelphia, Pennsylvania

Introduction

A Database Management System (DBMS) provides control of operational data. Database Systems have been used extensively for civil engineering applications on different computer platforms. Today's available Relational Database software (e. g., dBase III, FoxPro, Paradox) has simplified computer programming and made data maintenance and query possible, even for the computer novice.

Recently, the subject of Knowledge‑Based Expert System (KBES) has been attracting more and more attention in the civil engineering and construction management professions. Construction professionals, in particular, believe that Expert System technology could play an important role in improving the quality and productivity of construction. A number of Expert System Applications have been developed within recent years either with typical Artificial Intelligent Languages such as LISP and PROLOG, or with Expert System (ES) Environments (e.g., ART, KEE, Nexpert Object) and Shells (e.g., EXSYS, VP-EXPERT, Level 5 Object). With development levels ranging from very early stages to fully operational stages, the applications tend to provide, or are already providing, great assistance for the construction industry in the area of project management, construction methods, legal issues, and human-resource management.

In order to utilize the data stored in the DBMS, in conjunction with the KBES to provide decision making support according to the information input, the integration of these two system is necessary to bring the utmost benefit to the user.

Situation

In modern construction operations, it is becoming increasingly important that more consideration be given to the decision making process. The variety of information, available to solve the construction problems of today, determines that unique solution which will provide the best results under the anticipated conditions. A good or bad decision might directly affect the production rate and profitability of the construction project. There are many factors involved in the decision making process, such as: Internal Data (project information, organizational information, etc.), External Information (site conditions, capital availability, etc.), and Personnel Information (union relationship, labor availability, public relations, etc.). The magnitude of these data is tremendous, and thus requires the use of a database management system to support the decision making process. Meanwhile, substantial expert system application research and development have been made in the construction industry [S. Mohan, 1990]. The reason for this progress is the inherent variance resulting from construction's one-of-a-kind production technology, and from external influences such as weather, regulatory agencies and the like. Also, expertise in the construction field tends to be based upon individual experience which is passed down from one generation of engineers to the next. Thus, structured approaches to decision making are difficult to develop. However, the decision rules are very much like the "IF (condition) THEN (action)" format which is easily represented in a rule-based expert system [R. E. Levitt, 1987]. On the one hand, the novice or the field engineer needs to make decisions without the assistance of experienced personnel to provide suggestions, with the result being that decisions, that are based upon partial information or knowledge are deficient and negatively affect the profitability of the construction. On the other hand, even experienced experts may not remember all of the factors involved and thereby might neglect some small but very critical factors.

DBMS are now being used to manipulate all of the data related to a project while KBES's extracted from the field experience and published resources provide valuable suggestions with certain levels of confidence for the consideration of the decision maker. The DBMS being integrated with a KBES will be able to provide details, presently available in both systems, with fill functionalities. This kind of integrated system would certainly be of great help to the construction industry.

Previous research has shown either a lack of integration between KBES's and DBMS's, or only that complicate connections have been provided between knowledge-based system and databases. This is partially due to the software limitation at the time. State-of-the-art computer technology; however, provides the possibility of designing systems that can sufficiently utilize the advantages of both the KBES and DBMS. Such a system could also provide a friendly end user interface.

Previous Work

Many database management systems have been developed in the civil and construction engineering area in recent years. In 1987, F. S. Liou and M. Abullatif developed an integrated project management system model using dBase III. The database system provided the information for different construction management activities, such as: estimating, scheduling, procurement, and accounting. This system illustrate the advantages of database management through easy data transaction and readjustment of schedule and estimates. A. Tavaloli, J. J. Masehi, and C. S. Collyard developed an equipment management system (FLEET) in 1990. FLEET consists of inventory information, maintenance, cost and time, and report generator modules. This system was developed with dBase Relational Database Software and contributed to the efficiency and profitability of the contractors. In February 1992, A. C. Giannotti and D. J. Fisher submitted a paper entitled: "Project Information Management System - Another Approach". In this paper, they provided a mainframe database information system called PIMS, which receives information from project team members, inputs data and shares it with all users as soon as it is generated. PIMS offers the project manager valuable information as it occurs, with a minimal expenditure of time; and furthermore, it supports the management of information production, improves quality control and relieves technical personnel of tedious administrative and clerical duties. P. Shen and P. Udo-Inyang also developed a dBase III Management Information System for Job Cost Control on Construction Projects (1vBSJC) in September, 1992. In addition to these prototypes, there are many operational DBMS's which are being used in various fields of construction. For instance, the Pennsylvania Department of Transportation uses CMS (Contract Management System) and CDS (Construction Data System), which were both developed from dBase IV, to manipulate the data for accounting, procurement, payment, project status control, cost control, documentation and daily project data processing. A large number of DBMS's are also used elsewhere to fulfill the needs of the owner, architecture/engineer and contractor.

Early in 1986, S. J. Fenves provided people in construction with some basic ideas about expert systems in his paper entitled: "What is an Expert System?" Also that same year, C. N. Kostem discussed in his paper, "Attributes and Characteristics of Expert System", the key points that should be considered by the expert system developer. Then, G. A. Finn and K. F. Reinschmidt described efforts made by the Stone & Webster Engineering Corporation in the area of expert systems. Their applications involved different expert system shells and environments, which were applied to actual situations and which have provided great benefits for the user. Their applications have been distributed widely to several hundred users.

In 1990, S. Mohan did a survey on the existing applications of expert systems in the construction area. The survey results are provided in Table 1. Previous research relative to integrating a DBMS and a KBES include the following: In 1987, J. R. Chahine and B. N. Janson interfaced a database system with an expert system in their Retaining Wall Management Application. They used dBase III to develop a database environment containing structure and management data. The KBES, TOPSI, was called from the DBMS. The KBES, which was available at that time, was less flexible than dBase III and could not call any external program. This made the interfacing between the DBMS and the KBES only unilateral and restricted the functionality of the integration. In 1989, R D. Logcher, M. T. Wang, and F. H. Chen applied the KBES, Expert‑MCA, as query language for the mainframe database system CAPCES (for Construction Appropriations Programming Control and Execution System) developed with FOCUS DBMS. The system tried to focus on how information can be extracted from the data, and how knowledge can be accumulated from the extraction process. The integration provide the possibility of obtaining some useful information from an existing database. By coupling data stored in a database with heuristic knowledge, the value of both systems was increased. In April 1993, T. M. Adams provided an alternative method to apply expert systems technology. Adams set up the knowledge table (fact base) in the DBMS to represent knowledge, and used Structured Query Language (SQL) to act as the inference engine. The principle of this method is to store the knowledge as database relations between facts, concepts, and objects. However, this technique can only be applied to a very narrow scope of applications which satisfy unique conditions. In April 1993, C. K. Soh, A. K. Soh, and K. Y. Lai introduced their integrated knowledge-base-database model. The IPDOS (Interactive Preliminary Design of Offshore Structures) was developed with K-Base program embedded in a PC-based dBase III DBMS. The application makes it possible to apply expert system technology to an existing DBMS to configure the basic 3, 4, 6, and 8 legged platforms for routine oil- and gas-related functions and to provide recommendations according to the environmental and geographic condition, etc.

Database Management System

As defined by Hansen & Hansen [G. W. Hansen & J. V. Hansen, 1992], a database is a collection of inter-related data items that can be processed by one or more application systems, and a database management system (DBMS) represents systems software that facilitates the management of a database.

The early development of database technology started with file access methods. The development went through the random access stage, hierarchical stage, and network stage. In the 1970's, E. Codd's research revolutionized database development by introducing a model to access and manipulate data in terms of its logical characteristics. Presently, Relational Database systems dominate the commercial marketplace. By applying Object-Oriented model technology in Relational Database Systems, via the mechanical conversion from Object-Oriented Models to Relational Models, the complicated normalization procedures for the traditional model development can be avoided. In the near future, the Object-Oriented Database Systems will provide powerful ways for data manipulation which focus on data itself instead of the manipulating language. Refer to Figure 1.

The advantages of a database management system are:

Figure 1. Time Line Evaluation of a Database System [Hansen & Hansen, 1992]

To design a DBMS application, Data Base Design Life Cycle (DBDLC) normally contains three stages: Conceptual Design; Logical Design; and Physical Design. In the conceptual design stage, a simple approach would be to establish the Object-Oriented Model and then change it to a Relational Model (Schema: a model of a set of database states). The advantage of this procedure is that an Object Oriented Model intuitively reflects the user's world and by transfer, the resulting Schema would be automatically normalized. The basic mechanism is shown in Figure 2:

Knowledge-Based Expert System

Knowledge-Based Expert Systems have created much excitement in the civil engineering computer user community, similar to the emergence of FORTRAN in the 1950's, problem‑oriented languages in the 1960's, and CAD in the 1970's [S. J. Fenves,1986]. The typical features of today's construction environment show the need for an expert-system‑like technology for improving construction quality and productivity. These features are [Levitt, R E., 185, 1987]:,

Figure 2. A Generic Database Management System [J. P. Ignizio. 1991].

·        Experience dependent industry.

·        Heuristic approaches and incomplete information for decision making.

·        Computer illiteracy of many construction professionals and managers.

A Knowledge-Based Expert System (KBES) is formally defined as [Gasching, 1981]:

Knowledge-based expert systems are interactive computer programs incorporating judgement, experience, rules of thumb, intuition, and other expertise to provide knowledgeable advice about a variety of tasks.

The Expert System (ES) normally includes Knowledge Base, Inference Engine, Working Memory, and Rule Adjuster. The User and Knowledge Engineer access the ES by the Interface. The definitions for typical nomenclature are [J. P. Ignizio, 1991; G. A. Finn et al, 1986]:

Knowledge Base is the part of the ES in which the domain knowledge is stored as facts and heuristic rules.

Domain Knowledge is the knowledge related to a specific area of application, as opposed to general knowledge or common sense knowledge.

Inference Engine is also known as a knowledge processor. It serves as the knowledge processing element of an expert system to merge facts with rules in order to develop or to infer new rules.

Working Memory holds or stores the rules under consideration during the consultation session and stores the results of the inference process as new rules.

Rule Adjuster is the rule editor where the knowledge engineer enters rules into the knowledge base. The adjuster also checks the various rules.

Heuristic Rules are rules developed through intuition, experience, insight, and judgment. In a KBES, heuristic rules are used to reduce the number of paths searched in the inference network.

Rules-of-Thumb, are similar to heuristic rules and are usually empirical in nature, e.g., based on experience and intuition, with no mathematical or scientific proof.

A general overview of a KBES is shown in Figure 3. An expert system "shell" includes all of the components depicted in Figure 3, excluding the knowledge base. The knowledge engineer can use the shell to develop the knowledge base and then insert it into the architecture to form a complete expert system. This choice gives knowledge engineers the advantage that they can focus attention on the development of the knowledge base without repeatedly developing all the support elements of the knowledge base. The development of the expert system starts with problem identification. After the problem statement and description, the decision is made that the application of an expert system is necessary. From domain engineers and published resources, the knowledge acquisition and knowledge representation would be implemented and developed in a prototype. The reasoning part of the KBES includes forward chaining and backward chaining: Forward chaining is reasoning from facts to the conclusions resulting from these facts. Backward chaining involves reasoning in reverse from a hypothesis, a potential conclusion to be proved, to the facts which support the hypothesis. The elements of the development of an expert system are summarized in Figure 4.

One the main advantages of the employment of an expert system is achieved through the separation of the knowledge base (i.e., what we know) from the inference engine (i.e., what we do). The separation itself lends the shell to rule-base transparency, ease of maintenance, and the employment of expert shells. Though expert system shells provide less flexibility, it seems that the use of traditional AI language (LISP or PROLOG), in order to achieve additional flexibility, would be offset by the additional expenditure of time and funds. Shells allow the knowledge engineers to concentrate their attention on knowledge acquisition and knowledge representation. The system's ultimate performance depends on the knowledge that has been acquired and the manner in which it has been represented, and not the design of the inference engine. The development of shells enable people to apply expert system technology without extensive experience or skill for programming in AI lan­guage. This kind of development provides the possibility to match the prediction, originally proposed by Mohan in 1990 and stated below:

Many small expert systems and company proprietary expert system will be routinely used in industry. A good number of expert systems will be capable of interfacing with graphics, data-base-management pro­grams, and algorithmic programs.

The expert-system-building process will consolidate and document construction knowledge currently dis­persed among many individuals, journals, and books. This consolidation of domain knowledge will give rise to the building of large expert systems that will have frame-based knowledge representation with graphic input and output, and will be integrated with several external programs.

Integration

Traditional data processing generally involves more predictable and algorithmic transformations of information such as the calculation of costs for estimating. Knowledge processing involves knowledge transformations such as generalizations and abstractions. As extensive DBMS applications were developed, so too was the ability to manipulate large data structures. In addition to Management Information Systems, many Decision Support Systems were also developed to provide the support for management and decision making. In order to apply advanced expert system technology to decision support processes, according to information already stored in the DBMS, a challenge appears: How does one integrate a KBES into an existing DBMS? The rapid advancement in computer technology has enabled the computer to handle both knowledge and data processing simultaneously. However, state-of-the-art

DBMS and KBES software still have defects in data processing or knowledge processing, respectively.

In some KBES environment software, developers have started to integrate plain databases (tables or forms) into the system, but these internal databases are still not as powerful or as fully functional as a regular DBMS. Regardless, creators of existing DBMS applications find direct database, knowledge-base integration to be extremely useful and acceptable by most DBMS users. Previous solutions of embedding a KBES into a DBMS did not prove satisfactory, since integration limited the utilization of the KBES.

Recent research has summarized the categories of technology which are necessary to integrate knowledge-base and database. Basically, there are five approaches to system architecture as shown in Figure 5:

Figure 5 displays the differences in these approaches. Approaches (b) and (c) focus upon a Knowledge-Based Expert System and Approaches (d) and (e) focus upon a Database Management System. Approach (a) concentrates on the link between the DBMS and KBES.

The subject of integration is attracting more attention now because today's technology enables users to access data from multiple platforms. The original method of transferring files or data was to save the data into an ASCII file, then transfer it using communication software or mechanical means. This was especially true when a KBES was developed in an AI language and only run on Al machines. However, the exact data a knowledge system needs may not be known until the analysis is well underway. Furthermore, additional infusions of data are required as reasoning progresses. Therefore, the former method obviously would not be suitable for the field, because of the issues of time and skill required.

Since both the DBMS software and KBES shells are able to be used on a PC platform under the Microsoft Windows' environment, it is reasonable to consider utilizing Windows* as a integration between the DBMS and KBES. Dynamic Data Exchange (DDE) and Object Linking and Embedding (OLE) are able to transfer data from one Windows* software to another Windows software. Most of the state-of -the-art DBMS and KBES software are able to call an external program, using their own embedded programming language so as to execute one application from another.

The DDE and OLE methods proposed for the integration of a DBMS and a KBES are similar to Approach (a) of Figure 5. The mechanism of the integration is shown in Figure 6.

DDE is a Windows* protocol that enables one application to share data with other applications that behave according to the DDE protocol. Using DDE methods, users have access to data created and stored in another application and can also send commands and data to other applications. OLE is a Windows* protocol that provides access to the functions of another application without having to leave the client application and open the server application each time the user wants to make a change.

The DDE method allows two programs to engage in a "conversation", where the program that initiates the conversation is the "client", and the program that responds to the client is the "server". The client initiates the conversation, but may also send (poke) data to the server. The client may request data from the server, or may request the server to carry out a server-specific command such as printing a file or running a macro.

For example, using Level 5 Object KBES and Paradox for Windows DBMS, the integration implemented in Object Level 5 PRL language DDE Syntax resembles the followings:

The normal DDE conversation procedure in Level 5 Object is:

Instead of transferring data between applications using DDE, OLE provides the ability to store and display an entire object from another application. Through the DDE and OLE methods, the system can achieve a KBES and a DBMS integration.

Conclusions

Today, Personal Computers (PC's) are a particularly convenient and inexpensive means through which a DBMS and KBES may be developed and tested. For the construction industry, PC's provide great convenience for people in the field. Through relatively inexpensive networks and modems, users of PC's are able to communicate with each other very easily. The Windows° environment provides a widely employed user interface. The Windows environment also makes the integration of distinct systems easier and does not require advanced programming skills. The limitation of this integration is that certain software can only support DDE or OLE as a client, but not as a server.

Using the available technology, it is possible to develop an integrated system combining DBMS and KBES. The integrated system provides high transparency, through which the domain engineer can easily understand the rules and data. The advantage of this is obvious since the information feedback would need to be changed constantly for a prototype system. Even after implementation and validation, the changing world will require the system to incorporate new standards and verify unusual assumptions or decisions. After all, a computer itself will never replace the human expert. At its best in construction engineering, the computer will be able to assist the human expert to make decisions with its powerful "memorization" and "calculation" abilities.

References

Adams, T. M., "Knowledge Representation and Processing in Relational Data Base", Journal of Computer in Civil Engineering, Vol. 7, No. 2, April, 1993.

Adeli, H., Expert System in Construction and Structural Engineering, Chapman and Hall, New York, NY, 1988.

Alkass, S., and Harris, F., "Expert System for Earthmoving Equipment Selection in Road Construction", Journal of Construction Engineering and Management, Vol. 114, No. 3, September, 1988.

Alkass, S., and Aronian, A., "Computer‑Aided Equipment Selection for Concrete Placing", Concrete International, December, 1990. .

Alkass, S., and Aronian, A., and Moselhi, O., "Computer-Aided Equipment Selection for transporting and Placing Concrete", Journal of Construction Engineering and Management, Vol. 119, No. 3, September, 1993.

Allen, R H., Expert System for Civil Engineering: Knowledge Representation, ASCE, New York, NY, 1992.

Amirkhanian, S. N., and Baker, N. L., "Expert System for Equipment Selection for Earth-Moving Operation", Journal of Construction Engineering and Management, Vol. 118, No. 2, June, 1992.

Arciszewski, T., and Rossman, L. A., Knowledge Acquisition in Civil Engineering, ASCE, New York, NY, 1992.

Arockiasamy, M., "Retain Wall Design: RETWALL", Expert System Applications for Structural, Transportation, and Environmental Engineering, CRC Press, Boca Raton, FL, 1993.

Barton, P., Information Systems in Construction management: Principles and Applications, Batsford Academic and Education, London, UK,1985.

Chahine, J. R, and Janson, B. R, "Interfacing Data Bases with Expert Systems: A Retaining Wall management Application", Microcomputer in Civil Engineering, Vol. 2, No. 1, March, 1987.

Computer Technology in Construction, Thomas Telford, London, BK, 1985.

Diekmann. J. E., and Kraiem, Z., "Explanation of Construc­tion Engineering Knowledge in Expert System", Journal of Construction Engineering and Management, Vol. 114, No.3, September, 1988.

Fenves, S. J., "What is an Expert System", Expert in Civil Engineering, ASCE, New York, NY, 1986.

Finn, G. A., and Reinschmidt, K. F., "Expert System in an Engineering-Construction Firm", Expert in Civil Engineering, ASCE, New York, NY, 1986.

Garcia, O. N., and Chien, Y. T., Knowledge-Based System: Fundamental and Tools, IEEE Computer Society Press, Los Alsmitos, CA, 1991.

Giannotti, A. C., and Fisher, D. J., "Project Information Management System - Another Approach", Journal of management in Engineering, Vol. 9, No. 1, January, 1993.

Giarratano, J., and Riley, G., Expert Systems: Principles and Programming, PW S-KENT Publishing Company, Boston, MA, 1989.

Gibson Jr., G. E., and Bell, L. C., " Integrated Data-Base Systems", Journal of Construction Engineering and Management, Vol. 118, No. 1, March, 1992.

Hanna, A. S., and Willenbrock, J. H., and Sanvido, V. E., "Knowledge Acquisition and Development for Formwork Selection System", Journal of Construction Engineering and Management, Vol. 118, No. 1, March, 1992.

Ransen, G. W., and Hansen, J. V., Database Management and Design„ Prentice-Hall, Inc., Englewood Cliffs, NJ, 1992.

Hendrickson, C, and Au, T., Project Management for Construction, Prentice-Hall, Inc., Englewood Cliffs, NJ, 1989.

Ignizio, J. P., Introduction to Expert System: The Development and Implementation of Rule-Based Expert System, McGraw-Hill, Inc., New York, NY, 1991.

Kostem, C. N., "Attributes and Characteristics of Expert System", Expert in Civil Engineering, ASCE, New York, NY, 1986.

Level 5 Object for Microsoft Windows Reference Guide Release 3.0, Information Builders, Inc., New York, NY, 1993

Levitt, R E., "Expert Systems in Construction: State of the Art", Expert System for Civil Engineers: Technology and Application, M. L. Maher, ed., ASCE, New York, NY, 1987

Levitt, R. E., Research Areas For the Application of Knowledge Based Expert Systems to Construction Engineering and Management, White Paper for NSF workshop, Campaign, IL, 1985

Leu, S. S., Object-Oriented Representation Model of Construction Technology Information, Ph.D. dissertation, The University of Michigan, MI, 1992.

Liou, F. S., and Abullatif, M., "Database Management System for Construction Managers", 31st Annual Meeting of the American Association of Cost Engineering, AACE, Morgantown, WV, 1987.

Logcher, R. D., and Wang, M. T., and Chen, F. H. S., "Knowledge Processing for Construction Management Data Base", Journal of Construction Engineering and Management, Vol. 115, No. 2,1une, 1989.

Mathis, H. M., The Design and Development of an Expert System Construction Environment for Inclusion of Data Resource Knowledge, Ph.D. dissertation, Texas A&M University, Texas, 1989.

Mohan, S., "Expert System Applications in Construction Management and Engineering", Journal of Construction Engineering and Management, Vol. 116, No. 1, March, 1990_

Neil, J. M., Construction Cost Estimating for Project Control, Prentice-Hall, Inc., Englewood Cliffs, NJ, 1982

PARADOX for Windows User's Guide & OBJECTPAL Developer's Guide, Borland, Inc., Scotts Valley, CA, 1992.

Rumble Jr., J. R., and Smith, F. J., Database Systems in Science and Engineering, Adam Hilger, Inc., Bristol, 1990.

Shafer, S. L., and Westney, R. E., "Six Steps to Successful Expert System", 31st Annual Meeting of the American Association of Cost Engineering, AACE, Morgantown, WV, 1987.

Shen, P., Management Information System for Job Cost Control on Construction Projects. Technical Report No. CE/CM/MSE 92-01, Presented to the Department of Civil Engineering, Temple University, Philadelphia, in 1992, in partial fulfillment of the requirements for the degree of Master of Science in Engineering.

Soh, C. K., and Soh, A. K-, and Lai, K. Y., "Making Data Banks More Useful with Knowledge-Embedded Data‑Base Systems", Journal of Computing in Civil Engineering, Vol. 7, No. 2, April, 1993.

Udo-Inyang, P. D., and Shen, P., "Management by Exception for Construction Job Cost Control", Proceedings of the 29th Annual Conference, Associated Schools of Construction, Colorado State University, Fort Collins, Colorado