Understanding why machine learning algorithms may fail is usually the task of human expert that uses domain knowledge and contextual information to discover systematic shortcomings in either data or algorithm. In this paper, we propose a semantic referee, which able extract qualit ative features errors emerging from deep frameworks suggest corrections. The referee relies on ontological reasoning about spatial order characterize terms their relations with environment. Using semantics,...

We consider chordal RCC-8 networks and show that we can check their consistency by enforcing partial path with weak composition. prove this using the fact relations from maximal tractable subsets H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">8</sub> , C Q of have patchwork property. The use has important practical consequences demonstrate implementation new reasoner PyRCC∇, which is developed extending state art PyRCC8. Given an network only...

We improve the state-of-the-art method for compression of web and other similar graphs by introducing an elegant technique which further exploits clustering properties observed in these graphs. The analysis experimental evaluation our shows that it outperforms currently best Boldi et al. achieving a better ratio retrieval time. Our exhibits vast improvements on certain families graphs, such as social networks, taking advantage their compressibility characteristics, ensures will not worsen...

We improve the state-of-the-art method for checking consistency of large qualitative spatial networks that appear in Web Data by exploiting scale-free-like structure observed their constraint graphs. propose an implementation scheme triangulates graphs input and uses a hash table based adjacency list to efficiently represent reason with them. generate random using Barabási-Albert (BA) model preferential attachment mechanism. test our approach on already existing datasets have been...

We introduce and study the problem of obtaining a spatial or temporal configuration that maximizes number constraints satisfied in qualitative constraint network (QCN). call this MAX-QCN prove it is NP-hard for most calculi. also propose complete generic branch bound algorithm solving problem. This builds on techniques used literature consistency checking minimal labeling given QCN. In particular, we make use tractable subclass relations, chordal graph provided by triangulation input QCN,...

Qualitative Spatial &amp; Temporal Reasoning (QSTR) is a major field of study in Symbolic AI that deals with the representation and reasoning spatio- temporal information an abstract, human-like manner. We survey current status QSTR from viewpoint approaches, identify certain future challenges we think that, once overcome, will allow to meet demands adapt real-world, dynamic, time-critical applications highly active areas such as machine learning data mining.

We introduce, study, and evaluate a novel algorithm in the context of qualitative constraint-based spatial temporal reasoning, that is based on idea variable elimination, simple general exact inference approach probabilistic graphical models. Given constraint network N, our enforces particular directional local consistency which we denote by ←-consistency. Our discussion restricted to distributive subclasses relations, i.e., sets relations closed under converse, intersection, weak...

Abstract We survey the use and effect of decomposition-based techniques in qualitative spatial temporal constraint-based reasoning, clarify notions a tree decomposition, chordal graph, partitioning their implication with particular constraint property that has been extensively used literature, namely, patchwork. As consequence, we prove recently proposed approach was presented study by Nikolaou Koubarakis for checking satisfiability networks lacks soundness. Therefore, becomes quite...

There has been interest in recent literature tackling very large real world qualitative spatial networks, primarily because of the datasets that have been, and are to be, offered by Semantic Web community scale up millions nodes. The proposed techniques for such networks employ following two approaches retaining sparseness their underlying graphs reasoning with them: (i) graph triangulation sparse matrix implementation, (ii) partitioning parallelization. Regarding latter approach, an...

We introduce and study a notion of robustness in Qualitative Constraint Networks (QCNs), which are typically used to represent reason about abstract spatial temporal information. In particular, given QCN, we interested obtaining robust qualitative solution, or, scenario it, is satisfiable that has higher perturbation tolerance than any other, other words, more chances remain valid after it altered. This challenging problem requires consider the entire set scenarios whose size usually...

We introduce, study, and evaluate a novel algorithm in the context of qualitative constraint-based spatial temporal reasoning that is based on idea variable elimination, simple general exact inference approach probabilistic graphical models. Given constraint network [Formula: see text], our utilizes particular directional local consistency, which we denote by text]-consistency, order to efficiently decide satisfiability text]. Our discussion restricted distributive subclasses relations,...