Faculty of Electrical Engineering University of Belgrade CaseBased

Faculty of Electrical Engineering University of Belgrade Case-Based Reasoning Davitkov Miroslav, 2011/3116

1. Case-Based Reasoning definition • Case-Based reasoning (CBR), broadly construed, is the process of solving new problems based on the solutions of similar past problems. • CBR is reasoning by remembering: It is a starting point for new reasoning • Case-Based Reasoning is a well established research field that involves the investigation of theoretical foundations, system development and practical application building of experience-based problem solving. 2 / 25

1. Case-Based Reasoning definition Everyday examples of CBR : • An auto mechanic who fixes an engine by recalling another car that exhibited similar symptoms • A lawyer who advocates a particular outcome in a trial based on legal precedents or a judge who creates case law. • An engineer copying working elements of nature (practicing biomimicry), is treating nature as a database of solutions to problems. • Case-based reasoning is a prominent kind of analogy making. 3 / 25

2. CBR problem solver 1. Case – previously made and stored experience item 2. Case-Base – core of every case – based problem solver - collection of cases 4 / 25

2. CBR problem solver • A case-based problem solver solves new problems primarily by reuse of solutions from the cases in the case-base. • For this purpose, one or several relevant cases are selected. • One of the core assumptions behind CBR is that similar problems have similar solutions. 5 / 25

2. CBR problem solver • Once similar cases are selected, the solution(s) from the case(s) are adapted to become a solution of the current problem. • When a new (successful) solution to the new problem is found, a new experience is made, which can be stored in the case-base to increase its competence, thus implementing a learning behavior. 6 / 25

3. Types of CBR There are three main types of CBR that differ significantly from one another concerning case representation and reasoning: 1. Structural (a common structured vocabulary, i. e. an ontology) 2. Textual (cases are represented as free text, i. e. strings) 3. Conversational 4. (a case is represented through a list of questions that varies from one case to another ; knowledge is contained in customer / agent conversations) 7 / 25

4. CBR Cycle • Despite the many different appearances of CBR systems, the essentials of CBR are captured in a surprisingly simple and uniform process model. • The CBR cycle is proposed by Aamodt and Plaza. • The CBR cycle consists of 4 sequential steps around the knowledge of the CBR system. 8 / 25

4. CBR Cycle Problem New Case RETRIEVE Learned Case Retrieved Case Previous Cases RETAIN New Case General Knowledge Tested / Repaired Case Confirmed Solution REUSE REVISE Solved Case 9 / 25 Suggested Solution

4. CBR Cycle 4. 1. Retrieve • One or several cases from the case base are selected, based on the modeled similarity. • The retrieval task is defined as finding a small number of cases from the case-base with the highest similarity to the query. • This is a k-nearest-neighbor retrieval task considering a specific similarity function. • When the case base grows, the efficiency of retrieval decreases => methods that improve retrieval efficiency, e. g. specific index structures such as kd-trees, case-retrieval nets, or discrimination networks. 10 / 25

4. CBR Cycle 4. 2. Reuse • Reusing a retrieved solution can be quite simple if the solution is returned unchanged as the proposed solution for the new problem. • Adaptation (if required, e. g. for synthetic tasks). • Several techniques for adaptation in CBR - Transformational adaptation - Generative adaptation • Most practical CBR applications today try to avoid extensive adaptation for pragmatic reasons. 11 / 25

4. CBR Cycle 4. 3. Revise • In this phase, feedback related to the solution constructed so far is obtained. • This feedback can be given in the form of a correctness rating of the result or in the form of a manually corrected revised case. • The revised case or any other form of feedback enters the CBR system for its use in the subsequent retain phase. 12 / 25

4. CBR Cycle 4. 4. Retain • The retain phase is the learning phase of a CBR system (adding a revised case to the case base). • Explicit competence models have been developed that enable the selective retention of cases (because of the continuous increase of the case-base). • The revised case or any other form of feedback enters the CBR system for its use in the subsequent retain phase. 13 / 25

5. CBR and the Future Internet • The development of the future internet is affected by two major factors: semantics and collaboration. • Two of the most influencing developments of the Semantic Web are: - the resource description language RDF (Resource Description Framework) - the knowledge representation language OWL (Web Ontology Language), which is based on RDF • Already before the development of RDF and OWL, XML has been used as a case representation within the case-based reasoning community. 14 / 25

5. CBR and the Future Internet • There is a notable similarity between the ontologies developed within semantic applications and the representation of cases in structural case-based reasoning. • Due to this similarity RDF and OWL both lend themselves to be used as case representation languages and thus expand the possibilities of case-based reasoning within the general WWW. • There are technological and methodological similarities between ontologies and structured case-based reasoning and there are synergies that can be reached by merging both approaches. 15 / 25

5. CBR and the Future Internet • Case. ML - an RDF based Case Markup Language (by Chen and Wu); Case. ML offers a domain-independent case ontology and also aims to make case-based reasoning available within the Semantic Web. • SERVOGrid (by Aktas et al. ) – also uses RDF for case representation; It is embedded in a conversational case-based reasoning system that aids scientists in finding resources such as program code or data that are needed to solve a specific task by assisting them in describing the necessary resources using meta data. 16 / 25

5. CBR and the Future Internet • j. COLIBRI framework - OWL is being used as the case interchange language; It is planned to advance the already distributed framework towards an architecture consisting of Semantic Web Services (SWS) where problem solving methods are represented as Web Services; In order to use these services the whole case-based reasoning process is decomposed into single tasks, which are then carried out by according Web Services. 17 / 25

5. CBR and collaborative filtering • There is a close relation between collaborative filtering and CBR and these two can benefit from each other. • Example 1: Collaborative filtering is used to assess the similarity between songs in a CBR system creating custom music compilations (Co. A) [Aguzzoli et al. ]. • Example 2: A community based web search that uses the results of previous web searches of similar users in order to improve web search results [Briggs and Smyth]. 18 / 25

6. CBR applications • During the past twenty years, many CBR applications have been developed, ranging from prototypical applications build in research labs to large-scale fielded applications developed by commercial companies. • Application areas of CBR include: - help-desk and customer service - recommender systems in electronic commerce - knowledge and experience management - medical applications and applications in image processing - applications in law, technical diagnosis, design, planning - applications in the computer games and music domain. 19 / 25

7. CBR compared to other methods • We will compare CBR with the rule induction algorithm of machine learning. • Like a rule-induction algorithm, CBR starts with a set of cases or training examples; it forms generalizations of these examples, albeit implicit ones, by identifying commonalities between a retrieved case and the target problem. 20 / 25

7. CBR compared to other methods • The key difference, however, between the implicit generalization in CBR and the generalization in rule induction lies in when the generalization is made. • A rule-induction algorithm draws its generalizations from a set of training examples before the target problem is even known; that is, it performs eager generalization. • This is in contrast to CBR, which delays (implicit) generalization of its cases until testing time – a strategy of lazy generalization. • CBR therefore tends to be a good approach for rich, complex domains in which there are myriad ways to generalize a case. 21 / 25

8. Criticism of the CBR • Critics of CBR argue that it is an approach that accepts anecdotal evidence as its main operating principle. • Without statistically relevant data for backing and implicit generalization, there is no guarantee that the generalization is correct. • There is recent work that develops CBR within a statistical framework and formalizes case-based inference as a specific type of probabilistic inference; thus, it becomes possible to produce case-based predictions equipped with a certain level of confidence. 22 / 25

9. Conclusion • The number of CBR approaches and applications developed up to now has become quite large. • There is a significant number of CBR research groups and commercial companies, which develop CBR methods, software components, and applications on a regular basis. • CBR is not only a technology but also a (process oriented) method. • The combination of CBR with various other technologies within a great bandwidth of applications has become increasingly attractive for researchers as well as business professionals. 23 / 25

10. References • Ralph Bergmann, Klaus-Dieter Althoff, Mirjam Minor, Meike Reichle, Kerstin Bach: Case-Based Reasoning: Introduction and Recent Developments • Benjamin Heitmann, Conor Hayes: Enabling Case-Based Reasoning on the Web of Data • A. Aamodt, E. Plaza: Case-Based Reasoning: Foundational Issues, Methodological Variations, and System Approaches 24 / 25

Thank you for your attention! Questions? davitkov. miroslav@gmail. com dm 113116 m@student. etf. rs