One of the bottlenecks of current recognition (and graph matching) systems is their assumption of one-to-one feature (node) correspondence. This assumption breaks down in the generic object recognition task where, for example, a collection of features at one scale (in one image) may correspond to a single feature at a coarser scale (in the second image). Generic object recognition therefore req...

Semantic matching is a fundamental step in implementing data sharing applications. Most systems automating this task however limit themselves to finding simple (one-to-one) matching. In fact, complex (many-to-many) matching raises a far more difficult problem as the search space of concept combinations is often very large. This article presents Indigo, a system discovering complex matching in t...

In this paper, we address the problem of planning the textual organization of instructions. We take the view that natural language generation (NLG) is a mapping process of diierent levels of conceptual and textual representations. Within this framework , we consider the mapping between the text's semantic representation and its rhetorical structure. We argue that such a mapping is not direct, b...

A many-to-many k-disjoint path cover of a graph joining two disjoint vertex subsets S and T of equal size k is a set of k vertex-disjoint paths between S and T that altogether cover every vertex of the graph. It is classified as paired if each source in S is required to be paired with a specific sink in T , or unpaired otherwise. In this paper, we develop Ore-type sufficient conditions for the ...

This paper aims to introduce the issues related to the syntactic alignment of a dependency-based multilingual parallel treebank, ParTUT. Our approach to the task starts from a lexical mapping and then attempts to expand it using dependency relations. In developing the system, however, we realized that the only dependency relations between the individual nodes were not sufficient to overcome som...

This work presents a grapheme-based approach of English-to-Chinese (E2C) transliteration, which consists of many-to-many (M2M) alignment and conditional random fields (CRF) using accessor variety (AV) as an additional feature to approximate local context of source graphemes. Experiment results show that the AV of a given English named entity generally improves effectiveness of E2C transliteration.

The functioning of a multimodal integration system requires metadata, such as ontologies, that describe media resources and media components. Such metadata are generally application and domain dependent, which causes difficulties when media need to be shared across domains. Thus, there is a need for a mechanism that can relate the key terms and/or media components in data from different sources...

Learning Decompositional Shape Models from Examples Alex Levinshtein Master of Science Graduate Department of Computer Science University of Toronto 2005 We present an algorithm for automatically constructing a decompositional shape model from examples. Unlike current approaches to structural model acquisition, in which one-to-one correspondences among appearance-based features are used to cons...

We show that the full version of the so-called ‘rural hospital theorem’ (Roth, 1986) generalizes to many-to-many matching where agents on both sides of the market have separable and substitutable preferences.

We count the number of alignments of N ≥ 1 sequences when match-up types are from a specified set S ⊆ N . Equivalently, we count the number of nonnegative integer matrices whose rows sum to a given fixed vector and each of whose columns lie in S. We provide a new asymptotic formula for the case S = {(s1, . . . , sN ) | 1 ≤ si ≤ 2}.