Bisson, G. 1992a. Conceptual clustering in a first order logic representation. In Neumann, B., editor, Proc. ECAI--92: Tenth European Conference on Artificial Intelligence. John Wiley and Sons. 458--462.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... and decision making [19, 28] Conceptual clustering,by contrast, puts cluster representation in the foreground and searches for clusters that have good representations in a given description language [16, 37, 39] Examples of description languages include variants of predicate logic [7, 37] as well as probabilistic languages that list attribute probabilities [16, 41] Conceptual clustering has two important advantages in the framework of Figure 1. First, by deriving intelligible descriptions of clusters, it facilitates cluster interpretation. The importance of this point follows ....

G. Bisson. Conceptual Clustering in a First Order Logic Representation. In 10th European Conf. on Artificial Intelligence, pages 458--462, 1992.

....is a growing interest in first order learners, however existing proposals for distances between non ground atoms have some drawbacks. In this paper we develop a new measure for the distance between nonground atoms. 1 Introduction In learning systems based on clustering (e.g. C0.5 [3] KBG [1]) and in instance based learning (e.g. 9, ch.4] RIBL [6] a measure of the distance between objects is an essential component. Good measures exist for distances between objects in an attribute value representation (see e.g. 9, ch. 4] Recently there is a growing interest in using more ....

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, pages 458--462. John Wiley & Sons, 1992.

.... [21] Claudien [20] Probabilistic Relational Models [44] Cohen s Flipper (in [17] 60] and RDBC [43] e.g. Quinlan s Foil can also be considered an upgrade of either Michalski s AQ (1983) or CN2, RIBL upgrades the classical k nearest neighbor algorithm (using a first order distance due to [6]) SRT and Tilde upgrade the wellknown decision (and regression) tree paradigm incorporated in CART [12] and C4.5 [55, 56] Warmr upgrades Apriori [2, 1] Maccent upgrades the Maximum Entropy approach in [5] De Raedt and Dzeroski s PAC learning results (as well as its incorporation in the ....

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the Tenth European Conference on Artificial Intelligence, pages 458--462. John Wiley & Sons, 1992.

....The problem we study is the following: given some set X and a metric d on X , how can we extend d into a metric on the set of all (finite) subsets of X . Distances between composed objects and between sets of objects have applications in many domains such as cluster analysis (e.g. TIC [2] KBG [1]) computational geometry [8] machine learning (e.g. 9, ch.4] RIBL [6] Existing proposals for measures between point sets all have some problems: some are trivial and not very well suited for applications (e.g. the Hausdorff metric) others do not satisfy all the properties of ....

....computed in time bounded by a polynomial in A, B and T . 2 5 Normalised matching metric. Instance based learning systems such as RIBL [7] and clustering algorithms (e.g. agglomerative clustering algorithms using distances) make use of normalised similarity measures, i.e. measures in the interval [0,1]. In this section we develop a normalised distance between set of points based on a normalised distance between points. In some applications (e.g. algorithms for clustering where the distance between clusters shouldn t depend on the size of the objects) it is desirable to work with normalised ....

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the Tenth European Conference on Artificial Intelligence, pages 458--462. John Wiley & Sons, 1992.

....The conceptual clustering [12] 9] within the machine learning field represents the AI approach towards the automatic class inference. Unlike the conventional clustering methods, the conceptual ones put the emphasis on cluster intensional descriptions of various forms : conjunctive concepts [12, 2] within a logical formalisms, probabilistic concepts [9] or Bayesian clusters [5] Descriptions are even integrated into the clustering process : the search for homogeneous groups is guided by a description quality criterion which, in general, does not merely limit to proximity. Moreover, the ....

.... objects in the clustering process relies strongly on the existence of a hierarchical structure of classes (probabilistic concept trees in the sub sequent terminology) Interesting results on similarity based clustering on FOL descriptions of complex individuals have been first reported in [2] and further extended in [7] In [3] a description of a possible way to adapt the method on object based representations may be found. The solution suggested emphasizes the relational structure of an object based knowledge base but disregards the available class structures on different object ....

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, Vienna, Austria, pages 458--462, 1992.

....[73] and developed, for instance, by Lebowitz [74] Fisher et al. 75] and others. Because it is impossible to compute all possible generalizations, many authors use a similarity measure for objects as a guideline for generalization, for example, Lebowitz for attribute value descriptions, Bisson [19, 76] for logical representations, and the distance guided generalization for graphs, using MatchBox s results, which is described below. Another problem is that often a class of objects cannot be described by a single prototype because the class consists of several subclasses. A prototype has to be ....

G. Bisson. Conceptual clustering in a first order logic representation. In B. Neumann, editor, Proc. of the Tenth European Conference on Artificial Intelligence ECAI-92, pages 458--462. John Wiley & Sons, ltd., Chichester, 1992.

....of the respective flats. The flat dissimilarity depends, in turn, on the dissimilarity of the initial objects via owner. Both values depend recursively on themselves. A possible way to deal with such a deadlock is to compute the values as solutions of a system of linear equations (see [2]) 2 not to mix with component attributes A single system is composed for each strongly connected component that occurs in both networks. In the system, the variables x i correspond to pairs of objects which may be reached from the initial pair o, o 0 by the same sequences of relational links ....

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, Vienna, Austria, pages 458--462, 1992.

....with an unsupervised algorithm in order to explore interactions between supervised and unsupervised learning agents. To this end we are developing our own relational clustering algorithm which will permit unsupervised first order learning. Our approach is in some respects similar to that of KBG (Bisson, 1992) which also performs conceptual clustering over a first order logic representation. KBG finds similarities between the entities found in relations, whereas our algorithm attempts to find similarities between the relations that hold between entities. Our algorithm combines DINUS (Lavrac Dzeroski, ....

G. Bisson, Conceptual Clustering in a First Order Logic Representation, in Proceedings of Tenth European Conference on Artificial Intelligence (ECAI92), B. Neumann (Ed.), Wiley, 1992, 459-462.

....space of Cola 2, let us first describe roughly Sprite. 3.1 Sprite Sprite is a conceptual clustering tool in the system Mobal. In this paper Sprite can be regarded as re implementation of the conceptual clustering algorithm of the system Kbg 2 in Prolog. Therefore, we refer the reader to [Bisson 92a, Bisson 92b] for a detailed description of the similarity measure guided (polynomial) conceptual clustering approach used in Sprite. There are only three major difference between Kbg and Sprite. First, generalized class descriptions are not pruned in Sprite which is done in Kbg (see [Bisson 91] ....

Gilles Bisson. Conceptual Clustering in a First Order Logic Representation. In Proc. of the 10th ECAI, pp. 558--462, 1992.

....where concepts are defined by logical formulae [11] Others use probabilistic concepts [3] where a concept can be thought of as a prototype (a typical instance) along with a probability distribution on each dimension. More recent works extend clustering techniques to higher level languages: KBG [1] deals with first order logic. Kluster [6] uses a Kl one like language to avoid computationnal complexity and still keep comfortable representative power. However, there are domains where many numerical aspects are to be considered: in such domains, logic based formalisms apply difficultly, and ....

Bisson, G. (1992) Conceptual Clustering in a First Order Logic Representation. Proceedings of the Tenth European Conference on Artificial Intelligence, pp. 458--462. J. Wiley & Sons.

....The last step aims at building a hierarchical system of rules from the generalization graph. In this step, Kbg drops all premises in the class descriptions which are not neccessary to discriminate between the instances of different classes. For a detailed description of the learning step see [Bisson 92a, Bisson 92b] the rule construction step is described in [Bisson 91] 4.2 Construction of the Kbg input In research on conceptual clustering it is usually assumed that the descriptions of the objects for clustering is given as a conjunction of attribute value pairs. This assumption makes sense ....

Gilles Bisson. Conceptual Clustering in a First Order Logic Representation. In ECAI92P, pp. 558--462, 1992.

....230 A. Ketterlin, P. Gan carski and J.J. Korczak cepts are defined by logical formulae [MS83] Others use probabilistic concepts [Fis87] where a concept is defined by a probability distribution on each dimension. Other work extends clustering techniques to higher level languages: KBG [Bis92] deals with first order logic, and Kluster [KM94] uses a Kl one like language that mitigates computational complexity, while still retaining consider representational power. However, there are domains where numeric data must be considered. In such domains, logic based formalisms are difficult to ....

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the Tenth European Conference on Artificial Intelligence, pages 458--462. J. Wiley and Sons, 1992.

....an opposite position, by adapting algorithms to deal with the data they will meet. Early attempts to handle more complex formalisms include (Michalski Stepp 1983) where an extension of propositional logic is used. More recent works extend clustering techniques to higher level languages: KBG (Bisson 1992) deals with first order logic. Kluster (Kietz Morik 1994) uses a Kl one like language to avoid computational complexity and still keep comfortable representative power. However, two main characteristics of real world databases may make existing algorithms hard to apply. First, many domains ....

....of components. Since it does not use the composite clusters hierarchy as a component matcher , the process of adding an object to a cluster has a cost exponential in the number of components (all the possible bindings between components and a predefined list of attributes are tested) KBG (Bisson 1992) and Kluster (Kietz Morik 1994) both employ high level languages (respectively first order logic and description logic) Both systems build a DAG of clusters, instead of a hierarchy. But both work bottom up, and do not support incrementality. This may be prohibitive with large databases. ....

Bisson, G. 1992. Conceptual clustering in a first order logic representation. In Proceedings of the Tenth European Conference on Artificial Intelligence, 458-- 462. J. Wiley and Sons.

....relational descriptions as proposed by [8] and [24] for instance. Of very interest to the problems to be solved in bioinformatics are machine learning developments relying on clustering algorithms. Biological objects are systems of high complexity such that relational descriptions have to be used [3, 20]. Recently, first order clustering methods are introduced using distance information only [12, 4] The demands for interpreting the data being generated by genome sequencing are expanding more than ever. Machine learning methods may essentially support the analysis, interpretation, and prediction ....

G. Bisson. Conceptual clustering in a first order logic representation. In B. Neumann, editor, Proceedings of the 10th European Conference on Artificial Intelligence, pages 458--462. John Wiley & Sons, Ltd, 1992.

....the computation of the distances. For instance, the distance could be the Euclidean distance d 1 between the values of one or more numerical attributes, or it could be the distance d 2 as measured by a first order distance measure such as used in RIBL [ Emde and Wettschereck, 1996 ] or KBG [ Bisson, 1992 ] or [ Hutchinson, 1997 ] Given the distance at the level of the examples, the principles of instance based learning can be used to compute the prototypes. e.g. d 1 would result in a prototype function p 1 that would simply compute the mean for the cluster, whereas d 2 could result in function ....

....descriptions of the clusters through the representation of first order logical decision trees. 2. 4 PROBLEM SPECIFICATION By now we are able to formally specify the clustering problem: Given 2 Using Plotkin s [1970] notion of subsumption or the variants corresponding to structural matching [Bisson, 1992; De Raedt et al. 1997] ffl a set of examples E (each example is a set of tuples in a relational database or equivalently, a set of facts in Prolog) ffl a background theory B in the form of a Prolog program, ffl a distance measure d that computes the distance between two examples or ....

[Article contains additional citation context not shown here]

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, pages 458-- 462. John Wiley & Sons, 1992.

....is one of the fundamental unsupervised learning tasks, and has been intensively studied for propositional representations, both in statistics and in Machine Learning. For relational, first order representations, there has been a lot less research, but here also, existing work (e.g. 20] [1], 10] 3] has shown the application potential of clustering for tasks where class information is sparse, expensive to obtain, or unavailable. However, up to now, work in ILP has mostly concentrated on the task of conceptual clustering, i.e. restricting cluster formation to clusters with ....

....its pruning measure that selects a single level set of clusters from the induced clustering hierarchy. Results from empirical experiments are reported in section 4. In the related work section, we discuss other first order clustering systems, and in particular the relationship of RDBC to KBG [1], Cola 2 [10] and C0.5 [3] which have used distance functions within conceptual clustering. We conclude with a summary and some pointers to future work. 2 Distance Based Clustering For propositional representations, most clustering algorithms are based on elementary distance properties of the ....

[Article contains additional citation context not shown here]

G. Bisson. Conceptual clustering in a first order logic representation. In Proc. European Conference on Artificial Intelligence (ECAI-92), 1992.

....or to provide support to create a new one. This observation also applies to classification algorithms. No methodology or tool has been proposed to support the elaboration of conceptual clustering algorithms that build task specific ontologies. Work on conceptual clustering (e.g. 19] 8] 9] [2], 1] 26] has not been extensively applied to the problem of learning from corpora. One must however acknowledge that the application of conceptual clustering techniques to this domain is not straightforward, as existing algorithms must be previously adapted. As in the case of distances, the ....

Bisson G. 1992. Conceptual Clustering in a First Order Logic Representation. In Proceedings of 10th European Conference on Artificial Intelligence (ECAI'92), pp. 458-462, Vienna .

No context found.

BISSON G. 1992b. Conceptual Clustering in a First Order Logic Representation.

No context found.

Bisson, G. 1992a. Conceptual clustering in a first order logic representation. In Neumann, B., editor, Proc. ECAI--92: Tenth European Conference on Artificial Intelligence. John Wiley and Sons. 458--462.

No context found.

G. Bisson, `Conceptual clustering in a first order logic representation', in Proceedings of the 10th European Conference on Artificial Intelligence, Vienna, Austria, pp. 458--462, (1992).

No context found.

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, Vienna, Austria, pages 458--462, 1992.

No context found.

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, Vienna, Austria, pages 458--462, 1992.

No context found.

G. Bisson. Conceptual clustering in a first order logic representation. In ECAI-92, Vienne, Austria.

No context found.

G. Bisson. Conceptual clustering in a first order logic representation. In Proceedings of the 10th European Conference on Artificial Intelligence, Vienna, Austria, pages 458--462, 1992.

No context found.

Bisson, 1992. Conceptual clustering in a first-order logic representation, Proceedings of the 10 th ECAI, Vienna, pp. 458-462.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC