This thesis focuses on video watermarking robust against geometric attacks and video compressions. In addition to the requirements for an image watermarking algorithm, a digital video watermarking algorithm has to be robust against advanced video compressions, frame loss, frame swapping, aspect ratio change, frame rate change, intra- and inter-frame filtering, etc. Video compression, especially, the most efficient compression standard, H.264, and geometric attacks, such as rotation and cropping, frame aspect ratio change, and translation, are considered the most challenging attacks for video watermarking algorithms. In this thesis, we first review typical watermarking algorithms robust against geometric attacks and video compressions, and point out their advantages and disadvantages. Then, we propose our robust video watermarking algorithms against Rotation, Scaling and Translation (RST) attacks and MPEG-2 compression based on the logpolar mapping and the phase-only filtering method. Rotation or scaling transformation in the spatial domain results in vertical or horizontal shift in the log-polar mapping (LPM) of the magnitude of the Fourier spectrum of the target frame. Translation has no effect in this domain. This method is very robust to RST attacks and MPEG-2 compression. We also demonstrate that this method can be used as a RST parameters detector to work with other watermarking algorithms to improve their robustness to RST attacks. Furthermore, we propose a new video watermarking algorithm based on the 1D DFT (one-dimensional Discrete Fourier Transform) and 1D projection. This algorithm enhances the robustness to video compression and is able to resist the most advanced video compression, H.264. The 1D DFT for a video sequence along the temporal domain generates an ideal domain, in which the spatial information is still kept and the temporal information is obtained. With detailed analysis and calculation, we choose the frames with highest temporal frequencies to embed the fence-shaped watermark pattern in the Radon transform domain of the selected frames. The performance of the proposed algorithm is evaluated by video compression standards MPEG-2 and H.264; geometric attacks such as rotation, translation, and aspect-ratio changes; and other video processing. The most important advantages of this video watermarking algorithm are its simplicity, practicality and robustness.

The progressive adoption of XML by new communities of users has motivated the appearance of applications that require the management of large and complex collections, which present a large amount of heterogeneity. Some relevant examples are present in the fields of bioinformatics, cultural heritage, ontology management and geographic information systems, where heterogeneity is not only reflected in the textual content of documents, but also in the presence of rich structures which cannot be properly accounted for using fixed schema definitions. Current approaches for dealing with heterogeneous XML data are, however, mainly focused at the content level, whereas at the structural level only a limited amount of heterogeneity is tolerated; for instance, weakening the parent-child relationship between nodes into the ancestor-descendant relationship. The main objective of this thesis is devising new approaches for querying heterogeneous XML collections. This general objective has several implications: First, a collection can present different levels of heterogeneity in different granularity levels; this fact has a significant impact in the selection of specific approaches for handling, indexing and querying the collection. Therefore, several metrics are proposed for evaluating the level of heterogeneity at different levels, based on information-theoretical considerations. These metrics can be employed for characterizing collections, and clustering together those collections which present similar characteristics. Second, the high structural variability implies that query techniques based on exact tree matching, such as the standard XPath and XQuery languages, are not suitable for heterogeneous XML collections. As a consequence, approximate querying techniques based on similarity measures must be adopted. Within the thesis, we present a formal framework for the creation of similarity measures which is based on a study of the literature that shows that most approaches for approximate XML retrieval (i) are highly tailored to very specific problems and (ii) use similarity measures for ranking that can be expressed as ad-hoc combinations of a set of --basic' measures. Some examples of these widely used measures are tf-idf for textual information and several variations of edit distances. Our approach wraps these basic measures into generic, parametrizable components that can be combined into complex measures by exploiting the composite pattern, commonly used in Software Engineering. This approach also allows us to integrate seamlessly highly specific measures, such as protein-oriented matching functions.Finally, these measures are employed for the approximate retrieval of data in a context of highly structural heterogeneity, using a new approach based on the concepts of pattern and fragment. In our context, a pattern is a concise representations of the information needs of a user, and a fragment is a match of a pattern found in the database. A pattern consists of a set of tree-structured elements --- basically an XML subtree that is intended to be found in the database, but with a flexible semantics that is strongly dependent on a particular similarity measure. For example, depending on a particular measure, the particular hierarchy of elements, or the ordering of siblings, may or may not be deemed to be relevant when searching for occurrences in the database. Fragment matching, as a query primitive, can deal with a much higher degree of flexibility than existing approaches. In this thesis we provide exhaustive and top-k query algorithms. In the latter case, we adopt an approach that does not require the similarity measure to be monotonic, as all previous XML top-k algorithms (usually based on Fagin's algorithm) do. We also presents two extensions which are important in practical settings: a specification for the integration of the aforementioned techniques into XQuery, and a clustering algorithm that is useful to manage complex result sets.All of the algorithms have been implemented as part of ArHeX, a toolkit for the development of multi-similarity XML applications, which supports fragment-based queries through an extension of the XQuery language, and includes graphical tools for designing similarity measures and querying collections. We have used ArHeX to demonstrate the effectiveness of our approach using both synthetic and real data sets, in the context of a biomedical research project.

The explosive growth in the number of documents produced daily necessitates the development of effective alternatives to explore, analyze, and discover knowledge from documents. Document mining research work has emerged to devise automated means to discover and analyze useful information from documents. This work has been mainly concerned with constructing text representation models, developing distance measures to estimate similarities between documents, and utilizing that in mining processes such as document clustering, document classification, information retrieval, information filtering, and information extraction. <br /><br /> Conventional text representation methodologies consider documents as bags of words and ignore the meanings and ideas their authors want to convey. It is this deficiency that causes similarity measures to fail to perceive contextual similarity of text passages due to the variation of the words the passages contain, or at least perceive contextually dissimilar text passages as being similar because of the resemblance of words the passages have. <br /><br /> This thesis presents a new paradigm for mining documents by exploiting semantic information of their texts. A formal semantic representation of linguistic inputs is introduced and utilized to build a semantic representation scheme for documents. The representation scheme is constructed through accumulation of syntactic and semantic analysis outputs. A new distance measure is developed to determine the similarities between contents of documents. The measure is based on inexact matching of attributed trees. It involves the computation of all distinct similarity common sub-trees, and can be computed efficiently. It is believed that the proposed representation scheme along with the proposed similarity measure will enable more effective document mining processes. <br /><br /> The proposed techniques to mine documents were implemented as vital components in a mining system. A case study of semantic document clustering is presented to demonstrate the working and the efficacy of the framework. Experimental work is reported, and its results are presented and analyzed.

A <em>lexical analogy</em> is two pairs of words (<em>w</em>₁, <em>w</em>₂) and (<em>w</em>₃, <em>w</em>₄) such that the relation between <em>w</em>₁ and <em>w</em>₂ is identical or similar to the relation between <em>w</em>₃ and <em>w</em>₄. For example, (<em>abbreviation</em>, <em>word</em>) forms a lexical analogy with (<em>abstract</em>, <em>report</em>), because in both cases the former is a shortened version of the latter. Lexical analogies are of theoretic interest because they represent a second order similarity measure: <em>relational similarity</em>. Lexical analogies are also of practical importance in many applications, including text-understanding and learning ontological relations. <BR> <BR> This thesis presents a novel system that generates lexical analogies from a corpus of text documents. The system is motivated by a well-established theory of analogy-making, and views lexical analogy generation as a series of three processes: identifying pairs of words that are semantically related, finding clues to characterize their relations, and generating lexical analogies by matching pairs of words with similar relations. The system uses a <em>dependency grammar</em> to characterize semantic relations, and applies machine learning techniques to determine their similarities. Empirical evaluation shows that the system performs remarkably well, generating lexical analogies at a precision of over 90%.

Remotely-sensed hyperspectral imagery has become one the most advanced tools for analyzing the processes that shape the Earth and other planets. Effective, rapid analysis of high-volume, high-dimensional hyperspectral image data sets demands efficient, automated techniques to identify signatures of known materials in such imagery. In this thesis, we develop a framework for automatic material identification in hyperspectral imagery using adaptive similarity measures. We frame the material identification problem as a multiclass similarity-based classification problem, where our goal is to predict material labels for unlabeled target spectra based upon their similarities to source spectra with known material labels. As differences in capture conditions affect the spectral representations of materials, we divide the material identification problem into intra-domain (i.e., source and target spectra captured under identical conditions) and inter-domain (i.e., source and target spectra captured under different conditions) settings. The first component of this thesis develops adaptive similarity measures for intra-domain settings that measure the relevance of spectral features to the given classification task using small amounts of labeled data. We propose a technique based on multiclass Linear Discriminant Analysis (LDA) that combines several distinct similarity measures into a single hybrid measure capturing the strengths of each of the individual measures. We also provide a comparative survey of techniques for low-rank Mahalanobis metric learning, and demonstrate that regularized LDA yields competitive results to the state-of-the-art, at substantially lower computational cost. The second component of this thesis shifts the focus to inter-domain settings, and proposes a multiclass domain adaptation framework that reconciles systematic differences between spectra captured under similar, but not identical, conditions. Our framework computes a similarity-based mapping that captures structured, relative relationships between classes shared between source and target domains, allowing us apply a classifier trained using labeled source spectra to classify target spectra. We demonstrate improved domain adaptation accuracy in comparison to recently-proposed multitask learning and manifold alignment techniques in several case studies involving state-of-the-art synthetic and real-world hyperspectral imagery.

Background: Previous studies focused on customer loyalty and customer satisfaction. But no studies focused on the interaction between the financial advisor and their client. Therefore we have chosen to focus on this knowledge gap.Purpose: The purpose of this thesis is to illustrate how the interaction between financial advisors and customers affect financial advisory in investment decision making.Method: The thesis philosophy was positivistic because patterns were found with help of a survey. Furthermore is the paper quantitative since the thesis is measurable and it try to explain the interaction between the financial advisors and the clients demographic characteristics and how it influence the financial advice Conclusion: The theories thin-slicing and similarity attraction paradigm may not apply to the interaction between financial advisor and client. / Bakgrund:  Tidigare studier fokuserar på kundnöjdhet och kundlojalitet. Däremot saknas studier kring samspelet mellan finansiell rådgivare och kund. Därför har vi valt att fokusera på denna kunskapslucka Syfte: Syftet med uppsatsen är att belysa hur samspelet mellan finansiella rådgivare och kunder påverkar den finansiella rådgivningen vid ett investeringsbeslut. Metod: Uppsatsen har positivistisk undersökningsfilosofi då mönster hittades med hjälp av en undersökning. Vidare är uppsatsen kvantitativ eftersom den är mätbar och att den har undersökt om det finns några samband mellan den finansiella rådgivarens och kundens demografiska egenskaper samt om dessa påverkar rådgivningen. Slutsats: Teorierna thin-slicing och similarity attraction paradigm kan inte tillämpas i samspelet mellan finansiella rådgivare och kunder.

The explosive growth in the number of documents produced daily necessitates the development of effective alternatives to explore, analyze, and discover knowledge from documents. Document mining research work has emerged to devise automated means to discover and analyze useful information from documents. This work has been mainly concerned with constructing text representation models, developing distance measures to estimate similarities between documents, and utilizing that in mining processes such as document clustering, document classification, information retrieval, information filtering, and information extraction. <br /><br /> Conventional text representation methodologies consider documents as bags of words and ignore the meanings and ideas their authors want to convey. It is this deficiency that causes similarity measures to fail to perceive contextual similarity of text passages due to the variation of the words the passages contain, or at least perceive contextually dissimilar text passages as being similar because of the resemblance of words the passages have. <br /><br /> This thesis presents a new paradigm for mining documents by exploiting semantic information of their texts. A formal semantic representation of linguistic inputs is introduced and utilized to build a semantic representation scheme for documents. The representation scheme is constructed through accumulation of syntactic and semantic analysis outputs. A new distance measure is developed to determine the similarities between contents of documents. The measure is based on inexact matching of attributed trees. It involves the computation of all distinct similarity common sub-trees, and can be computed efficiently. It is believed that the proposed representation scheme along with the proposed similarity measure will enable more effective document mining processes. <br /><br /> The proposed techniques to mine documents were implemented as vital components in a mining system. A case study of semantic document clustering is presented to demonstrate the working and the efficacy of the framework. Experimental work is reported, and its results are presented and analyzed.

A <em>lexical analogy</em> is two pairs of words (<em>w</em>₁, <em>w</em>₂) and (<em>w</em>₃, <em>w</em>₄) such that the relation between <em>w</em>₁ and <em>w</em>₂ is identical or similar to the relation between <em>w</em>₃ and <em>w</em>₄. For example, (<em>abbreviation</em>, <em>word</em>) forms a lexical analogy with (<em>abstract</em>, <em>report</em>), because in both cases the former is a shortened version of the latter. Lexical analogies are of theoretic interest because they represent a second order similarity measure: <em>relational similarity</em>. Lexical analogies are also of practical importance in many applications, including text-understanding and learning ontological relations. <BR> <BR> This thesis presents a novel system that generates lexical analogies from a corpus of text documents. The system is motivated by a well-established theory of analogy-making, and views lexical analogy generation as a series of three processes: identifying pairs of words that are semantically related, finding clues to characterize their relations, and generating lexical analogies by matching pairs of words with similar relations. The system uses a <em>dependency grammar</em> to characterize semantic relations, and applies machine learning techniques to determine their similarities. Empirical evaluation shows that the system performs remarkably well, generating lexical analogies at a precision of over 90%.

This thesis has two related goals: the first involves the concept of self-similarity of images. Image self-similarity is important because it forms the basis for many imaging techniques such as non-local means denoising and fractal image coding. Research so far has been focused largely on self-similarity in the pixel domain. That is, examining how well different regions in an image mimic each other. Also, most works so far concerning self-similarity have utilized only the mean squared error (MSE). In this thesis, self-similarity is examined in terms of the pixel and wavelet representations of images. In each of these domains, two ways of measuring similarity are considered: the MSE and a relatively new measurement of image fidelity called the Structural Similarity (SSIM) Index. We show that the MSE and SSIM Index give very different answers to the question of how self-similar images really are. The second goal of this thesis involves non-local image processing. First, a generalization of the well known non-local means denoising algorithm is proposed and examined. The groundwork for this generalization is set by the aforementioned results on image self-similarity with respect to the MSE. This new method is then extended to the wavelet representation of images. Experimental results are given to illustrate the applications of these new ideas.

Since its introduction in 2004, the Structural Similarity (SSIM) index has gained widespread popularity as an image quality assessment measure. SSIM is currently recognized to be one of the most powerful methods of assessing the visual closeness of images. That being said, the Mean Squared Error (MSE), which performs very poorly from a perceptual point of view, still remains the most common optimization criterion in image processing applications because of its relative simplicity along with a number of other properties that are deemed important. In this thesis, some necessary tools to assist in the design of SSIM-optimal algorithms are developed. This work combines theoretical developments with experimental research and practical algorithms. The description of the mathematical properties of the SSIM index represents the principal theoretical achievement in this thesis. Indeed, it is demonstrated how the SSIM index can be transformed into a distance metric. Local convexity, quasi-convexity, symmetries and invariance properties are also proved. The study of the SSIM index is also generalized to a family of metrics called normalized (or M-relative) metrics. Various analytical techniques for different kinds of SSIM-based optimization are then devised. For example, the best approximation according to the SSIM is described for orthogonal and redundant basis sets. SSIM-geodesic paths with arclength parameterization are also traced between images. Finally, formulas for SSIM-optimal point estimators are obtained. On the experimental side of the research, the structural self-similarity of images is studied. This leads to the confirmation of the hypothesis that the main source of self-similarity of images lies in their regions of low variance. On the practical side, an implementation of local statistical tests on the image residual is proposed for the assessment of denoised images. Also, heuristic estimations of the SSIM index and the MSE are developed. The research performed in this thesis should lead to the development of state-of-the-art image denoising algorithms. A better comprehension of the mathematical properties of the SSIM index represents another step toward the replacement of the MSE with SSIM in image processing applications.