|The Proposal Details|
| Search Engine Tool based on Case Based Reasoning (CBR)
|The proposal is to make a search engine tool based on the Artificial Intelligence technique of Case Based Reasoning (CBR). The tool will provide a framework for implementing a CBR application. CBR systems provide a mechanism to record parameters that influence a decision and also the decision itself in the form of information units called "cases". A CBR system also has retrieval algorithms that retrieve an existing case if it is similar to a new situation and propose the same action / decision be taken as was done earlier.
It will be a problem resolution tool designed to help organizations to harness knowledge or expertise, share harnessed knowledge or expertise via the web, throughout the enterprise as well as with customers and partners. Knowledge can be refined and updated anytime from anywhere and information can be delivered across multiple channels, over the web, via email, phone and desktop.
|The tool will be AI based Case Based Reasoning solution that allows harnessing, refining, organizing and maintaining organizational knowledge or expertise through a rich set of knowledge modelers. The tool will provide a meta-model for representing knowledge into a repository called the case base. It will also provide a set of sophisticated matching algorithms that will allow search over the case base to locate information even when the inputs are not very precise.
Furthermore, the presence of sophisticated algorithms to search over volumes of knowledge, mechanisms to add new knowledge, modify existing knowledge in the light of new experience as well as recall the knowledge most relevant will lead to greater efficiencies in exception handling. Therefore the tool will be a perfect cognitive knowledge management and reuse vehicle.
In case-based reasoning (CBR) systems expertise is embodied in a library of past cases, rather than being encoded in classical rules. Each case typically contains a description of the problem, plus a solution and/or the outcome. The knowledge and reasoning process used by an expert to solve the problem is not recorded, but is implicit in the solution.
To solve a current problem, the problem is matched against the cases in the case base, and similar cases are retrieved. The retrieved cases are used to suggest a solution which is reused and tested for success. If necessary, the solution is then revised. Finally the current problem and the final solution are retained as part of a new case.
All case-based reasoning methods have in common the following process:
1.retrieve the most similar case (or cases) comparing the case to the library of past cases;
2.reuse the retrieved case to try to solve the current problem;
3.revise and adapt the proposed solution if necessary;
4.retain the final solution as part of a new case.
There are a variety of different methods for organizing, retrieving, utilizing and indexing the knowledge retained in past cases.
Retrieving a case starts with a (possibly partial) problem description and ends when a best matching case has been found. The subtasks involve:
1.identifying a set of relevant problem descriptors;
2.matching the case and returning a set of sufficiently similar cases (given a similarity threshold of some kind); and
3.selecting the best case from the set of cases returned.
Some systems retrieve cases based largely on superficial syntactic similarities among problem descriptors, while advanced systems use semantic similarities.
Reusing the retrieved case solution in the context of the new case focuses on: identifying the differences between the retrieved and the current case; and identifying the part of a retrieved case which can be transferred to the new case. Generally the solution of the retrieved case is transferred to the new case directly as its solution case.
Revising the case solution generated by the reuse process is necessary when the solution proves incorrect. This provides an opportunity to learn from failure.
Retaining the case is the process of incorporating whatever is useful from the new case into the case library. This involves deciding what information to retain and in what form to retain it; how to index the case for future retrieval; and integrating the new case into the case library.
A CBR tool should support the four main processes of CBR: retrieval, reuse, revision and retention. A good tool should support a variety of retrieval mechanisms and allow them to be mixed when necessary. In addition, the tool should be able to handle large case libraries with retrieval time increasing linearly (at worst) with the number of cases.
The case-base development methodology (CBDM):
Knowledge within the tool will be gained over 2 phases called Seeding Phase and the Incremental Phase. The Seeding Phase is the initial period of converting existing knowledge in whatever form, explicit or implicit into cases. Once the Seeding Phase is complete, the system can be used for knowledge based call resolution. Subsequently, knowledge is added to the system incrementally when a problem is encountered for which no solution exists. This phase of operation is called the Incremental Phase.
The case-base development methodology or CBDM will guide knowledge authors in converting existing documentation, fault logs, diagnostic records into well formed case knowledge during the Seeding Phase. The CBDM provides a clear set of steps as well as structures called case templates that enable the existing knowledge in different forms to be converted into reusable cases.
Knowledge Centric Helpdesks: Organizations are increasingly realizing that building a guided call resolution capability can substantially cut down support costs. Helpdesk environments traditionally generate reams of data on call resolution scenarios. A mechanism that captures and harvests this data can effectively smoother the delivery of support. An important measure of the call resolution capability of the helpdesk is the number of calls attended to without the need for escalation. The guided call resolution capability of the tool seeks to maximize resolution at this level.
Moreover, as enterprise processes mature, behavior gets more predictable. The tool may harvest this call resolution data for re-use. The ability to identify cases based on past experience means faster resolution of queries. The knowledge seeding facility of the tool will capture information that might otherwise be lost to a transient IT support workforce. This will let organizations leveraging their in-house support operators into creative knowledge providers who will handle only the non-repetitive tasks. The tool can be set up to engage the caller in a âconversationâ, asking pointed questions to narrow the scope of possible solutions from a universal set developed through âseedingâ process.
Experience based decision making:As enterprise processes mature, behavior gets more and more predictable since the processes become well-defined or standardized. However, there are situations that are deviations from the anticipated behavior and occur on a regular basis for a variety of reasons. For instance, a bank might offer a loan at a lower-than-normal rate in the light of competition. This will not be done as a rule but will be a one-off instance. Such deviations are called expectations to the otherwise standard process.
Expectations are a part of most processes â whether mature or at a fledging stage. These are unpredictable and hence the most common practice is to handle them via experience. There could be several parameters that could drive the decision that needs to be made in an exception situation. These parameters are implicitly understood by the Human mind and the personnel that have this understanding are called âexpertsâ. Different techniques have been tried out over time in an effort to elicit and capture this know-how in some form that can be reused. Doing so would reduce dependency on individuals and also make room for automation.
A technique that has been found to work well for exception handling is Case Based Reasoning (CBR). CBR systems provide a mechanism to record parameters that influence a decision and also the decision itself in the form of information units called âcasesâ. A CBR system also has retrieval algorithms that retrieve an existing case if it is similar to a new situation and propose the same action / decision be taken as was done earlier.
Often there is confusion on whether a case based solution works better or a rule based solution. The answer is not either or but often could be both. Case based solutions work better when:
The domain is not well understood or defined. Like in the instance of claims processing, there could be new types coming in every now and then.
The number of parameters involved in decision-making is large. In such situations, it is practically impossible for any expert to encode a rule.
The retrieval is not necessarily exact and even somewhat âcloseâ claims from history are required to be considered.
This tool can be used to handle fraudulent claims, a case would comprise of claim details as the âsituationâ and the action taken manually as the âdecisionâ. The decision could be as simple as âapproveâ or ârejectâ. Claim details will include all parameters that are evaluated to determine whether the claim is valid or fraudulent. The advantages of this tool in this scenario will be:
1.The cases are captured automatically based on what manually happens. So when a new kind of claim will arrive, the tool will simply capture the details and not propose any action. Once it is manually verified, the decision will also be recorded by the tool.
2.When a similar claim will arrive, the existing claims will be checked for any similarity. If found similar, the same action / decision will be proposed.
3.If the tool proposes a wrong decision, there can be feedback loop incorporated to âturn offâ the cases that were erring in decision making.
Diagnosis: Case-based diagnosis systems try to retrieve past cases whose symptom lists are similar in nature to that of the new case and suggest diagnoses based on the best matching retrieved cases. The majority of installed systems are of this type and there are many medical CBR diagnostic systems.
1.Assessment: case-based systems are used to determine values for variables by comparing it to the known value of something similar. Assessment tasks are quite common in the finance and marketing domains.
2.Design:Systems to support human designers in architectural and industrial design have been developed. These systems assist the user in only one part of the design process, that of retrieving past cases, and would need to be combined with other forms of reasoning to support the full design process.
Aamodt, A., (1989) Towards robust expert systems that learn from experience - an architectural framework. In John Boose, Brian Gaines, Jean-Gabriel Ganascia (eds.): EKAW-89; Third European Knowledge Acquisition for Knowledge-Based Systems Workshop , Paris, July 1989. pp 311-326.
Aamodt, A. (1991). A knowledge-intensive approach to problem solving and sustained learning, Ph.D. dissertation, University of Trondheim, Norwegian Institute of Technology, May 1991. (University Microfilms PUB 92-08460)
Aamodt, A. (1993) Explanation-driven retrieval, reuse, and learning of cases, In EWCBR-93: First European Workshop on Case-Based Reasoning. University of Kaiserslautern SEKI Report SR-93-12 (SFB 314) (Kaiserslautern, Germany, 1993) 279-284.
Althoff, K.D (1989). Knowledge acquisition in the domain of CNC machine centers; the MOLTKE approach. In John Boose, Brian Gaines, Jean-Gabriel Ganascia (eds.): EKAW-89; Third European Workshop on Knowledge-Based Systems, Paris, July 1989. pp 180-195.
Althoff, K-D. (1992). Machine learning and knowledge acquisition in a computational architecture for fault diagnosis in engineering systems. Proceedings of the ML-92 Workshop on Computational Architectures for Machine Learning and Knowledge Acquisition. Aberdeen, Scotland, July 1992.
Anderson, J. R., (1983). The architecture of cognition. Harvard University Press, Cambridge.
Aha, D. , Kibler, D. , and Albert, M. K. (1991). Instance-Based Learning Algorithms. Machine Learning, vol.6 (1).
Allemange, D. (1993) (Review of EWCBR-93, AICom, this issue).
Bareiss, R. (1989). Exemplar-based knowledge acquisition: A unified approach to concept representation, classification, and learning. Boston, Academic Press.