As Building Information Modeling (BIM) applications become more sophisticated and used within other knowledge domains, the limitations of existing data exchange and sharing methods become apparent. The integration of BIM and Geographic Information System (GIS) can offer substantial benefits to manage the planning process during the design and construction stages. Currently, building (and geospatial) data are shared between BIM software tools through a common data format, such as Industry Foundation Classes (IFC). Because of the diversity and complexity of domain knowledge across BIM and GIS systems, however, these syntactic approaches are not capable of overcoming semantic heterogeneity. This study uses semantic web technology to ensure the highest level of interoperability between existing BIM and GIS tools. The proposed approach is composed of three main steps; ontology construction, semantic integration through interoperable data formats and standards, and query of heterogeneous information sources. Because no application ontology is available to encompass all IFC classes with different attributes, we first develop an IFC-compliant ontology describing the hierarchy structure of BIM objects. Then, we can translate the building's elements and GIS data into semantic web standard formats. Once the information has been gathered from different sources and transformed into an appropriate semantic web format, the SPARQL query language is used in the last step to retrieve this information from a dataset. The completeness of the methodology is validated through a case study and two use case examples.

This thesis is concerned with the semantic function of linguistic elements which do not seem to contribute to the truth conditions of an utterance, that is, with 'non-truth-conditional' linguistic devices. The first part of the thesis is devoted to theoretical considerations, while the second part concentrates on 'concessive' linguistic devices, which form a sub-class of 'non-truth-conditional' expressions. The first chapter outlines the way in which traditional semantic theories have employed the notion of truth conditions to capture linguistic meaning and a series of problems with this approach are pointed out. The chapter ends with an overview of 'non-truth-conditional' linguistic devices. Chapter 2 is concerned with ways in which fundamentally truth-conditional theories of linguistic semantics have attempted to accommodate such expressions in their frameworks. In chapter 3, the discussion focuses on Argumentation Theory, which does not just accommodate non-truth-conditional meaning but, ultimately, treats all linguistic meaning in non-truth-conditional terms and leads to the untenable conclusion that the general intuition that utterances can give information about the world is an illusion. This is followed by a chapter devoted to Sperber & Wilson's cognitive Relevance Theory. It is argued that this theory offers an ideal framework for a semantic analysis of 'truth-conditional' and 'non-truth-conditional' expressions alike, while avoiding the problems encountered by other theories. The next three chapters investigate the nature of linguistic 'concessivity' and provide a critical survey of existing analyses of three specific 'concessive' devices: but, although, and even if. In each case, an original relevance-theoretic analysis in procedural terms is proposed. In the last chapter, the possibility of purely pragmatic (that is, unencoded) 'concessive' interpretations is explored, and, finally, the role of the concept of 'truth-conditional content' in a theory of utterance interpretation is reassessed.

A growing number of applications require dictionaries of words belonging to semantic classes present in specialized domains. Manually constructed knowledge bases often do not provide sufficient coverage of specialized vocabulary and require substantial effort to build and keep up-to-date. In this thesis, we propose a semi-supervised approach to the construction of domain-specific semantic lexicons based on the distributional similarity hypothesis. Our method starts with a small set of seed words representing the target class and an unannotated text corpus. It locates instances of seed words in the text and generates lexical patterns from their contexts; these patterns in turn extract more words/phrases that belong to the semantic category in an iterative manner. This bootstrapping process can be continued until the output lexicon reaches the desired size. We explore employing techniques such as learning lexicons for multiple semantic classes at the same time and using feedback from competing lexicons to increase the learning precision. Evaluated for extraction of dish names and subjective adjectives from a corpus of restaurant reviews, our approach demonstrates great flexibility in learning various word classes, and also performance improvements over state of the art bootstrapping and distributional similarity techniques for the extraction of semantically similar words. Its shallow lexical patterns also prove to perform superior to syntactic patterns in capturing the semantic class of words.

As a new topic in text mining, novelty detection is a natural extension of information retrieval systems, or search engines. Aiming at refining raw search results by filtering out old news and saving only the novel messages, it saves modern people from the nightmare of information overload. One of the difficulties in novelty detection is the inherent ambiguity of language, which is the carrier of information. Among the sources of ambiguity, synonymy proves to be a notable factor. To address this issue, previous studies mainly employed WordNet, a lexical database which can be perceived as a thesaurus. Rather than borrowing a dictionary, we proposed a statistical approach employing Latent Semantic Indexing (LSI) to learn semantic relationship automatically with the help of language resources. To apply LSI which involves matrix factorization, an immediate problem is that the dataset in novelty detection is dynamic and changing constantly. As an imitation of real-world scenario, texts are ranked in chronological order and examined one by one. Each text is only compared with those having appeared earlier, while later ones remain unknown. As a result, the data matrix starts as a one-row vector representing the first report, and has a new row added at the bottom every time we read a new document. Such a changing dataset makes it hard to employ matrix methods directly. Although LSI has long been acknowledged as an effective text mining method when considering semantic structure, it has never been used in novelty detection, nor have other statistical treatments. We tried to change this situation by introducing external text source to build the latent semantic space, onto which the incoming news vectors were projected. We used the Reuters-21578 dataset and the TREC data as sources of latent semantic information. Topics were divided into years and types in order to take the differences between them into account. Results showed that LSI, though very effective in traditional information retrieval tasks, had only a slight improvement to the performances for some data types. The extent of improvement depended on the similarity between news data and external information. A probing into the co-occurrence matrix attributed such a limited performance to the unique features of microblogs. Their short sentence lengths and restricted dictionary made it very hard to recover and exploit latent semantic information via traditional data structure.

The aim of this study was to investigate two factors that have been claimed to be associated with the word retrieval difficulties experienced by individuals with aphasia. These factors were aphasic subjects’ level of phonological awareness and their lexical semantic processing. (For complete abstract open document)

The role of the basal ganglia in human language function remains unknown, despite a corpus of literature documenting the association between vascular lesions of the dominant nonthalamic subcortical (NS) region and disordered language. Theories of subcortical language function have been postulated (e.g., Crosson, 1985; Wallesch & Papagno, 1988), however, research in this field has remained largely data-driven, providing limited descriptions of individuals with vascular NS lesions in terms of performance on standard off-line language measures. This approach has failed to reveal the underlying nature of these language deficits “locally” in terms of various dynamic and temporally constrained linguistic and nonlinguistic component processes. The current series of studies are based largely on the premise that such empirical data has the potential to speak more directly to the cogency of current theories proposing a subcortical role in language or related cognitive functions. The present thesis investigated the performance of individuals with dominant chronic vascular NS lesions, compared to matched control subjects, individuals with Parkinson’s disease (PD) (also assumed to have NS dysfunction) and subjects with cortical lesions (CL), on a series of experiments which allowed for the real-time examination of language processing, manipulating the degree to which automatic and attentional/strategic processing is invoked. The theoretical underpinning of these experiments hinges primarily on the proposed role of frontal-subcortical systems in mediating aspects of language via attentional/strategic mechanisms. Accordingly, it was hypothesised that the locus of impairment for individuals with NS lesions would be centred selectively on those facets of language processing which require increased recourse to these proposed frontal-subcortical cognitive capacities. The language abilities of 15 subjects with chronic dominant NS lesions, 15 matched control subjects, 14 matched subjects with CL, and 12 matched individuals with PD were examined initially on the Western Aphasia Battery (WAB) and the Boston Naming Test (BNT). Most NS subjects were classified as non-aphasic according to the WAB, however, circumscribed deficits were evidenced, typically in generative and confrontation naming. In contrast, the CL group showed significant deficits on most aspects of the WAB compared to matched normal control subjects, and presented with a more severe impairment than NS subjects overall on the WAB and in confrontation naming and repetition. The PD group performance was not significantly different from the matched control group, while PD subjects performed better than the NS group overall on the WAB. The same cohort of NS, CL, PD, and control subjects undertook a battery of complex language measures designed to place a range of higher-order cognitive demands on the language processing system. This battery included subtests from the Test of Language Competence-Expanded Edition (TLC-E), the Test of Word Knowledge (TOWK), and The Word Test-Revised (TWT-R). The NS, CL, and PD subjects presented with marked disturbances in those tasks involving cognitive-linguistic flexibility, sentence formulation, indeterminacy of meaning, and metalinguistic manipulation of the lexical-semantic system. Collectively, the off-line results suggest that those aspects of language processing which are more heavily reliant on higher-order cognitive capacities are selectively compromised in subjects with NS lesions and PD. This assumption was further examined and substantiated in a series of on-line lexical ambiguity priming experiments performed by a subset of the original NS subjects (n = 10), matched control subjects (n = 10), matched CL subjects (n=10), and matched PD subjects (n = 10). When lexical ambiguities were presented in a single word context as word triplets, NS subjects showed rapid nonselective lexical activation, suggesting that intact lexical-semantic information could be accessed via automatic routines, similar to control subjects. Unlike control subjects, however, NS subjects were unable to sustain any form of significant activation, implying a selective impairment in the ability to manipulate lexical-semantic information through attentional/controlled processing. This breakdown was qualitatively different to the controlled processing disturbance evidenced by CL subjects, who maintained nonselective meaning facilitation over time, while PD subjects showed a pattern of selective priming consistent with a reduction in attentional processing. The emerging picture of a dissociation between intact automatic processing and compromised attentional/strategic lexical processing in the NS subjects was further elucidated in an experiment examining the processing of unequibiased lexical ambiguities in isolation. In this study, NS and PD subjects showed rapid nonselective meaning facilitation, again implying intact automatic lexical processing. While control and CL subjects evidenced multiple meaning activation followed by selective facilitation of the dominant meaning, NS and PD subjects were unable to achieve selective meaning facilitation, instead showing a protracted period of nonselective lexical activation. This finding suggested that when ambiguities were encountered in isolation, there was not an absolute breakdown in attentional processing per se, but rather a circumscribed deficit in the selective attentional engagement of the semantic network on the basis of meaning frequency, possibly implicating a disturbance of inhibitory mechanisms within the semantic network. A cross-modal priming experiment was used to investigate how lexical ambiguities were processed and resolved in a biased sentential context. Initially, lexical activation for the neurological patient groups appeared influenced by contextual information to a greater extent than in normal controls, which may indicate delayed lexical decision making or disturbed automatic lexical access. Only the PD and NS individuals failed to then maintain selective facilitation of the contextually appropriate meaning, suggesting a breakdown in the attention-based control and maintenance of semantic activation on the basis of integrated sentential constraints. This finding was extended in another cross-modal priming experiment, where NS and PD subjects appeared unable to use discourse-level information to select meanings and develop topical inferences via attentional/strategic mechanisms, while CL subjects showed a selective disturbance of inference development. The results of this thesis have served to delineate certain dynamic aspects of language processing in individuals with NS lesions in terms of automatic lexical processing components and processes involving the attentional/strategic selection of meaning on the basis of meaning frequency and various types of contextual information. In general, the NS group showed a demarcation between intact automatic processing and a breakdown in attentional/strategic processing which was manifest differently depending on the conditions under which processing was invoked. Furthermore, the performance of NS subjects on attentional operations was able to be dissociated under certain conditions from CL group performance and was similar to the PD group’s performance in certain instances. These preliminary findings are consistent with recent theories proposing a role for frontal-subcortical systems in the “top-down” modulation of semantic processing via executive attentional and strategic mechanisms. Although a disturbance in these systems provides a parsimonious explanation of the NS and PD group performance, such conclusions are drawn tentatively with the caveat that the precise neuropathological basis of cognitive-linguistic deficits in these individuals remains unclear at present.

Thesis (Ph.D.)--University of Texas at Dallas, 2007. / Includes vita. Includes bibliographical references (leaves 137-146)

Zugl.: Zürich, Univ., Diplomarbeit, 2004

University, Diss.--Leipzig, 2007.

Zugl.: Stuttgart, Univ., Diss., 2005. / Druckausg. bei Jost-Jetter, Heimsheim erschienen.