Farhi Marir

Case-based reasoning: A review

Case-Based Reasoning (CBR) is a relatively recent problem solving technique that is attracting increasing attention. However, the number of people with first-hand theoretical or practical experience of CBR is still small. The main objective of this review is to provide a comprehensive overview of the subject to people new to CBR. The paper outlines the development of CBR in the US in the 1980s. It describes the fundamental techniques of CBR and contrasts its approach to that of model-based reasoning systems. 1 A critical review of currently available CBR software tools is followed by descriptions of CBR applications both from academic research and, in more detail, three CBR systems that are presently being used commercially. Each of the three commercial case studies highlights features that made CBR particularly suitable for the application. Moreover, the last case study describes a development methodology for implementing CBR systems. The paper concludes with a research agenda for CBR. A detailed categorized bibliography of CBR research is provided in a companion paper (Marir & Watson, 1994).

Case-based reasoning: a categorized bibliography

Case-Based Reasoning (CBR) is a fresh reasoning paradigm for the design of expert systems in domains that may not be appropriate for other reasoning paradigms such as model-based reasoning. As a result of this, and because of its resemblance to human reasoning, CBR has attracted increasing interest both from those experienced in developing expert systems and from novices. Although CBR is a relatively new discipline, there are an increasing number of papers and books being published on the subject. In this context, this bibliographic categorization is an accompanying paper to a review of CBR by the same authors. The objective of this paper is to help researchers quickly identify relevant references.

Developing a virtual reality interface for an integrated project database environment

Explores the potential of using virtual reality (VR) as an interface for an integrated project database using the World Wide Web. VRML (Virtual Reality Modelling Language), which is a Web-based standard, is explored as a means of remotely interrogating information stored within an integrated database. VRML allows the creation of 3D views and worlds which can be explored in real time. This paper describes the steps taken to link the OSCON integrated database to a VRML environment. It addresses the issue of using VR as the medium for the remote accessing of information within an integrated environment. For instance, site engineers can query the database from their sites if they have access to a modem and the Internet. Technologies are now available which can be used to this end. The Internet and its facilities should be exploited for the benefits of better management and retrieval of construction information. VR has often been looked on as a visualisation tool. This paper puts the argument that VR should be used as a user interface. For instance, the user should interact with a 3D column in VR rather than a column in a traditional database environment. This allows the construction practitioners better access to information which will motivate them to use integrated databases. In addition, VR should be used as a vehicle for classifying information. The user should be able to look at the design, costing, time planning and facilities management views according to his/her requirements. This may prove helpful in projecting information from a project integrated database.

Representing and Indexing Building Refurbishment Cases for Multiple Retrieval of Adaptable Pieces of Cases

CBRefurb is a case-based reasoning (CBR) system for the strategic cost estimation for building refurbishment. This domain is characterised by many uncertainties and variation. Its cost estimation involves large amount of interrelated factors whose impact is difficult to assess. This paper report on the problems faced by the building cost information Services (BCIS) databases and several rule-based expert systems to tackle this complex cost estimation problem and, the design and evaluation of CBRefurb system implemented using ReMind Shell. CBRefurb imitates the domain expert in its approach of breaking down the whole building work into smaller work (building items) by organising the refurbishment cases as a hierarchical structure composed of cases and subcases. The process of estimation imitate the expert by considering only these pieces of previous cases of similar situation (or context). For this purpose, CBRefurb defines some of the building and its component (or items) features as a global context and local context information used to classify cases and subcases into context cases and subcases, and to decompose the cost estimation problem into adaptable subproblems. This is followed by a two indexing schemes to suit the hierarchical structure of the case and the problem decomposition and to allow classification and retrieval of contextual cases. CBRefurb features consolidate the aim of the project that is allowing multiple retrieval of appropriate pieces of the refurbishment which are easier to adapt, reflecting the expert method of estimating cost for complex refurbishment work.

An enhanced grouping algorithm for vertical partitioning problem in DDBs

Distribution design involves making decisions on the fragmentation and allocation of data across the sites of a computer network. Vertical partitioning is the process of subdividing the attributes of a relation to generate fragments. In this paper, we propose an enhancement to our previous work for vertical partitioning algorithm using grouping approach. This algorithm starts from the attribute affinity matrix and generates initial groups based on the affinity values between attributes. Then, it attempts to merge the initial groups to produce final groups that will represent the fragments.

An adapted framework and case-based reasoning for business process redesign

This paper presents a new approach for business processes redesign (BPR). The approach uses case-based reasoning techniques to model past business redesign into a case or stories based on a developed BPR framework and best practices rules. The resulting case base of past failing or successful experiences are used to support business process designer and practitioner in redesigning or improving existing business processes. This work came to address the limitation of most methodologies developed to support the application of BPR principles in providing clear guidance on deriving a process design threatening the success of BPR.

A case-based approach to construction process activity specification

This paper investigates and proposes an original approach in specifying construction project process activities by making use of case-based reasoning (CBR) techniques. The work presented falls into the scope of the EPSRC funded COMMIT and CBRefurb projects. The COMMIT project aims at providing information management support in the context of distributed collaborative construction engineering. A brief summary of the problems encountered in project process specification is given. A new approach, based on CBR, is proposed. The CBR techniques and cycle are described. A subset of the COMMIT information management model, along with its implementation, dealing with project lifecycle description is presented. The paper then addresses the process activity case representation, indexing, retrieval and adaptation. Finally, the CIMM process CBR interface is presented.

Rhetorical structure theory for content-based indexing and retrieval of Web documents

The amount of information available on the Internet is currently growing at an incredible rate. However, the lack of efficient indexing is still a major barrier to effective information retrieval on the Web. This paper presents the design of a technique for content-based indexing and retrieval of relevant documents from a large collection of documents such as the Internet. The technique aims at improving the quality of retrieval by capturing the semantics of the documents. It introduces a thematic relationship between parts of text using a linguistics theory called rhetorical structure theory (RST) based on cue phrases to determine the set of rhetorical relations. Once these structures are determined, they can be saved into a database. We can then query that collection using not only keywords, as traditional information retrieval systems, but also rhetorical relations. The indexing and retrieval technique described in this paper is under development and initial results on a small number of documents have been very successful.

SEMIR: Semantic Indexing and Retrieving Web document using Rhetorical Structure Theory

The amount of information available on the Internet is currently growing at an incredible rate. The lack of efficient indexing is, however, still a major barrier to effective information retrieval of documents from the web. This paper proposes the design of a technique for content-based indexing and retrieval of relevant documents from a large collection of documents such as the Internet. The technique aims at improving the quality of retrieval by capturing the semantics of these documents. It performs the conventional keyword-based indexing and introduces a thematic relationship between parts of text using a linguistics theory called Rhetorical Structure Theory (RST). The indexing and retrieval technique described in this paper is currently under development, and initial results obtained from a small number of documents has been found to be more satisfactory, compared to those obtained using keyword-based indexing and retrieval techniques we have been using.

Developing a conceptual model for an integrated database

The paper describes a conceptual model developed by the OSCON team at the University of Salford in the UK in order to facilitate the development of an integrated database. A suite of demonstration applications is being developed These will consist of a CAD package, a cost estimating tool and an import/export link to an existing project planning package and a VRML interface. The software is being developed in C++ on the PC under Windows NT using the Microsoft Foundation Classes application framework. The ObjectStore OODBMS is being used in conjunction with the Object Engineering Workbench (OEW) modelling software to implement the project database. The OEW software is used to develop an object model which can support the applications which are within the scope of the project. The paper describes the conceptual core model and the methodology which is used to develop it. The conceptual model consists of a set of core objects linked by relationships which reflect the requirements of the integrated database.