People have complex thoughts, and they often express their thoughts with complex sentences using natural languages. This complexity may facilitate efficient communications among the audience with the same knowledge base. But on the other hand, for a different or new audience this composition becomes cumbersome to understand and analyze. Analysis of such compositions using syntactic or semantic measures is a challenging job and defines the base step for natural language processing. In this dissertation I explore and propose a number of new techniques to analyze and visualize the syntactic and semantic patterns of unstructured English texts. The syntactic analysis is done through a proposed visualization technique which categorizes and compares different English compositions based on their different reading complexity metrics. For the semantic analysis I use Latent Semantic Analysis (LSA) to analyze the hidden patterns in complex compositions. I have used this technique to analyze comments from a social visualization web site for detecting the irrelevant ones (e.g., spam). The patterns of collaborations are also studied through statistical analysis. Word sense disambiguation is used to figure out the correct sense of a word in a sentence or composition. Using textual similarity measure, based on the different word similarity measures and word sense disambiguation on collaborative text snippets from social collaborative environment, reveals a direction to untie the knots of complex hidden patterns of collaboration.

Im Vergleich zu einer stichwortbasierten Suche ermöglicht die semantische Suche ein präziseres und anspruchsvolleres Durchsuchen von (Web)-Dokumenten, weil durch die explizite Semantik Mehrdeutigkeiten von natürlicher Sprache vermieden und semantische Beziehungen in das Suchergebnis einbezogen werden können. Eine semantische, Entitäten-basierte Suche geht von einer Anfrage mit festgelegter Bedeutung aus und liefert nur Dokumente, die mit dieser Entität annotiert sind als Suchergebnis. Die wichtigste Voraussetzung für eine Entitäten-zentrierte Suche stellt die Annotation der Dokumente im Archiv mit Entitäten und Kategorien dar. Textuelle Informationen werden analysiert und mit den entsprechenden Entitäten und Kategorien versehen, um den Inhalt semantisch erschließen zu können. Eine manuelle Annotation erfordert Domänenwissen und ist sehr zeitaufwendig. Die semantische Annotation von Videodokumenten erfordert besondere Aufmerksamkeit, da inhaltsbasierte Metadaten von Videos aus verschiedenen Quellen stammen, verschiedene Eigenschaften und Zuverlässigkeiten besitzen und daher nicht wie Fließtext behandelt werden können. Die vorliegende Arbeit stellt einen semantischen Analyseprozess für Video-Metadaten vor. Die Eigenschaften der verschiedenen Metadatentypen werden analysiert und ein Konfidenzwert ermittelt. Dieser Wert spiegelt die Korrektheit und die wahrscheinliche Mehrdeutigkeit eines Metadatums wieder. Beginnend mit dem Metadatum mit dem höchsten Konfidenzwert wird der Analyseprozess innerhalb eines Kontexts in absteigender Reihenfolge des Konfidenzwerts durchgeführt. Die bereits analysierten Metadaten dienen als Referenzpunkt für die weiteren Analysen. So kann eine möglichst korrekte Analyse der heterogen strukturierten Daten eines Kontexts sichergestellt werden. Am Ende der Analyse eines Metadatums wird die für den Kontext relevanteste Entität aus einer Liste von Kandidaten identifiziert - das Metadatum wird disambiguiert. Hierfür wurden verschiedene Disambiguierungsalgorithmen entwickelt, die Beschreibungstexte und semantische Beziehungen der Entitätenkandidaten zum gegebenen Kontext in Betracht ziehen. Der Kontext für die Disambiguierung wird für jedes Metadatum anhand der Eigenschaften und Konfidenzwerte zusammengestellt. Der vorgestellte Analyseprozess ist an zwei Hypothesen angelehnt: Um die Analyseergebnisse verbessern zu können, sollten die Metadaten eines Kontexts in absteigender Reihenfolge ihres Konfidenzwertes verarbeitet werden und die Kontextgrenzen von Videometadaten sollten durch Segmentgrenzen definiert werden, um möglichst Kontexte mit kohärentem Inhalt zu erhalten. Durch ausführliche Evaluationen konnten die gestellten Hypothesen bestätigt werden. Der Analyseprozess wurden gegen mehrere State-of-the-Art Methoden verglichen und erzielt verbesserte Ergebnisse in Bezug auf Recall und Precision, besonders für Metadaten, die aus weniger zuverlässigen Quellen stammen. Der Analyseprozess ist Teil eines Videoanalyse-Frameworks und wurde bereits erfolgreich in verschiedenen Projekten eingesetzt. / The Semantic Web provides information contained in the World Wide Web as machine-readable facts. In comparison to a keyword-based inquiry, semantic search enables a more sophisticated exploration of web documents. By clarifying the meaning behind entities, search results are more precise and the semantics simultaneously enable an exploration of semantic relationships. However, unlike keyword searches, a semantic entity-focused search requires that web documents are annotated with semantic representations of common words and named entities. Manual semantic annotation of (web) documents is time-consuming; in response, automatic annotation services have emerged in recent years. These annotation services take continuous text as input, detect important key terms and named entities and annotate them with semantic entities contained in widely used semantic knowledge bases, such as Freebase or DBpedia. Metadata of video documents require special attention. Semantic analysis approaches for continuous text cannot be applied, because information of a context in video documents originates from multiple sources possessing different reliabilities and characteristics. This thesis presents a semantic analysis approach consisting of a context model and a disambiguation algorithm for video metadata. The context model takes into account the characteristics of video metadata and derives a confidence value for each metadata item. The confidence value represents the level of correctness and ambiguity of the textual information of the metadata item. The lower the ambiguity and the higher the prospective correctness, the higher the confidence value. The metadata items derived from the video metadata are analyzed in a specific order from high to low confidence level. Previously analyzed metadata are used as reference points in the context for subsequent disambiguation. The contextually most relevant entity is identified by means of descriptive texts and semantic relationships to the context. The context is created dynamically for each metadata item, taking into account the confidence value and other characteristics. The proposed semantic analysis follows two hypotheses: metadata items of a context should be processed in descendent order of their confidence value, and the metadata that pertains to a context should be limited by content-based segmentation boundaries. The evaluation results support the proposed hypotheses and show increased recall and precision for annotated entities, especially for metadata that originates from sources with low reliability. The algorithms have been evaluated against several state-of-the-art annotation approaches. The presented semantic analysis process is integrated into a video analysis framework and has been successfully applied in several projects for the purpose of semantic video exploration of videos.

Made available in DSpace on 2016-03-15T19:37:43Z (GMT). No. of bitstreams: 1 Ana Carolina Espirito Santo Lima.pdf: 2485278 bytes, checksum: 9843b9f756f82c023af6a2ee291f2b1d (MD5) Previous issue date: 2012-12-14 / Fundação de Amparo a Pesquisa do Estado de São Paulo / Social media have become a way of expressing collective interests. People are motivated by the sharing of information and the feedback from friends and colleagues. Among the many social media tools available, the Twitter microblog is gaining popularity as a platform for in-stantaneous communication. Millions of messages are generated daily, from over 100 million users, about the most varied subjects. As it is a rapid communication platform, this microblog spurred a phenomenon called television storytellers, where surfers comment on what they watch on TV while the programs are being transmitted. The Social TV emerged from this integration between social media and television. The amount of data generated on the TV shows is a rich material for data analysis. Broadcasters may use such information to improve their programs and increase interaction with their audience. Among the main challenges in social media data analysis there is sentiment analysis (to determine the polarity of a text, for instance, positive or negative), and sense disambiguation (to determine the right context of polysemic words). This dissertation aims to use machine learning techniques to create a tool to support Social TV, contributing specifically to the automation of sentiment analysis and disambiguation of Twitter messages. / As mídias sociais são uma forma de expressão dos interesses coletivos, as pessoas gostam de compartilhar informações e sentem-se valorizadas por causa disso. Entre as mídias sociais o microblog Twitter vem ganhando popularidade como uma plataforma para comunicação ins-tantânea. São milhões de mensagens geradas todos os dias, por cerca de 100 milhões de usuá-rios, carregadas dos mais diversos assuntos. Por ser uma plataforma de comunicação rápida esse microblog estimulou um fenômeno denominado narradores televisivos, em que os inter-nautas comentam sobre o que assistem na TV no momento em que é transmitido. Dessa inte-gração entre as mídias sociais e a televisão emergiu a TV Social. A quantidade de dados gera-dos sobre os programas de TV formam um rico material para análise de dados. Emissoras podem usar tais informações para aperfeiçoar seus programas e aumentar a interação com seu público. Dentre os principais desafios da análise de dados de mídias sociais encontram-se a análise de sentimento (determinação de polaridade em um texto, por exemplo, positivo ou negativo) e a desambiguação de sentido (determinação do contexto correto de palavras polis-sêmicas). Essa dissertação tem como objetivo usar técnicas de aprendizagem de máquina para a criação de uma ferramenta de apoio à TV Social com contribuições na automatização dos processos de análise de sentimento e desambiguação de sentido de mensagens postadas no Twitter.

A comunicação multilíngue é uma tarefa cada vez mais imperativa no cenário atual de grande disseminação de informações em diversas línguas. Nesse contexto, são de grande relevância os sistemas de tradução automática, que auxiliam tal comunicação, automatizando-a. Apesar de ser uma área de pesquisa bastante antiga, a Tradução Automática ainda apresenta muitos problemas. Um dos principais problemas é a ambigüidade lexical, ou seja, a necessidade de escolha de uma palavra, na língua alvo, para traduzir uma palavra da língua fonte quando há várias opções de tradução. Esse problema se mostra ainda mais complexo quando são identificadas apenas variações de sentido nas opções de tradução. Ele é denominado, nesse caso, \"ambigüidade lexical de sentido\". Várias abordagens têm sido propostas para a desambiguação lexical de sentido, mas elas são, em geral, monolíngues (para o inglês) e independentes de aplicação. Além disso, apresentam limitações no que diz respeito às fontes de conhecimento que podem ser exploradas. Em se tratando da língua portuguesa, em especial, não há pesquisas significativas voltadas para a resolução desse problema. O objetivo deste trabalho é a proposta e desenvolvimento de uma nova abordagem de desambiguação lexical de sentido, voltada especificamente para a tradução automática, que segue uma metodologia híbrida (baseada em conhecimento e em córpus) e utiliza um formalismo relacional para a representação de vários tipos de conhecimentos e de exemplos de desambiguação, por meio da técnica de Programação Lógica Indutiva. Experimentos diversos mostraram que a abordagem proposta supera abordagens alternativas para a desambiguação multilíngue e apresenta desempenho superior ou comparável ao do estado da arte em desambiguação monolíngue. Adicionalmente, tal abordagem se mostrou efetiva como mecanismo auxiliar para a escolha lexical na tradução automática estatística / Crosslingual communication has become a very imperative task in the current scenario with the increasing amount of information dissemination in several languages. In this context, machine translation systems, which can facilitate such communication by providing automatic translations, are of great importance. Although research in Machine Translation dates back to the 1950\'s, the area still has many problems. One of the main problems is that of lexical ambiguity, that is, the need for lexical choice when translating a source language word that has several translation options in the target language. This problem is even more complex when only sense variations are found in the translation options, a problem named \"sense ambiguity\". Several approaches have been proposed for word sense disambiguation, but they are in general monolingual (for English) and application-independent. Moreover, they have limitations regarding the types of knowledge sources that can be exploited. Particularly, there is no significant research aiming to word sense disambiguation involving Portuguese. The goal of this PhD work is the proposal and development of a novel approach for word sense disambiguation which is specifically designed for machine translation, follows a hybrid methodology (knowledge and corpus-based), and employs a relational formalism to represent various kinds of knowledge sources and disambiguation examples, by using Inductive Logic Programming. Several experiments have shown that the proposed approach overcomes alternative approaches in multilingual disambiguation and achieves higher or comparable results to the state of the art in monolingual disambiguation. Additionally, the approach has shown to effectively assist lexical choice in a statistical machine translation system

No description available.

We describe the automatic acquisition of a semantic network in which over 7,500 of the most frequently occurring nouns in the English language are linked to their semantically related concepts in the WordNet noun ontology. Relatedness between nouns is discovered automatically from lexical co-occurrence in Wikipedia texts using a novel adaptation of an information theoretic inspired measure. Our algorithm then capitalizes on salient sense clustering among these semantic associates to automatically disambiguate them to their corresponding WordNet noun senses (i.e., concepts). The resultant concept-to-concept associations, stemming from 7,593 target nouns, with 17,104 distinct senses among them, constitute a large-scale semantic network with 208,832 undirected edges between related concepts. Our work can thus be conceived of as augmenting the WordNet noun ontology with RelatedTo links. The network, which we refer to as the Szumlanski-Gomez Network (SGN), has been subjected to a variety of evaluative measures, including manual inspection by human judges and quantitative comparison to gold standard data for semantic relatedness measurements. We have also evaluated the network’s performance in an applied setting on a word sense disambiguation (WSD) task in which the network served as a knowledge source for established graph-based spreading activation algorithms, and have shown: a) the network is competitive with WordNet when used as a stand-alone knowledge source for WSD, b) combining our network with WordNet achieves disambiguation results that exceed the performance of either resource individually, and c) our network outperforms a similar resource, WordNet++ (Ponzetto & Navigli, 2010), that has been automatically derived from annotations in the Wikipedia corpus. iii Finally, we present a study on human perceptions of relatedness. In our study, we elicited quantitative evaluations of semantic relatedness from human subjects using a variation of the classical methodology that Rubenstein and Goodenough (1965) employed to investigate human perceptions of semantic similarity. Judgments from individual subjects in our study exhibit high average correlation to the elicited relatedness means using leave-one-out sampling (r = 0.77, σ = 0.09, N = 73), although not as high as average human correlation in previous studies of similarity judgments, for which Resnik (1995) established an upper bound of r = 0.90 (σ = 0.07, N = 10). These results suggest that human perceptions of relatedness are less strictly constrained than evaluations of similarity, and establish a clearer expectation for what constitutes human-like performance by a computational measure of semantic relatedness. We also contrast the performance of a variety of similarity and relatedness measures on our dataset to their performance on similarity norms and introduce our own dataset as a supplementary evaluative standard for relatedness measures.

No description available.

Semantic distance is a measure of how close or distant in meaning two units of language are. A large number of important natural language problems, including machine translation and word sense disambiguation, can be viewed as semantic distance problems. The two dominant approaches to estimating semantic distance are the WordNet-based semantic measures and the corpus-based distributional measures. In this thesis, I compare them, both qualitatively and quantitatively, and identify the limitations of each. This thesis argues that estimating semantic distance is essentially a property of concepts (rather than words) and that two concepts are semantically close if they occur in similar contexts. Instead of identifying the co-occurrence (distributional) profiles of words (distributional hypothesis), I argue that distributional profiles of concepts (DPCs) can be used to infer the semantic properties of concepts and indeed to estimate semantic distance more accurately. I propose a new hybrid approach to calculating semantic distance that combines corpus statistics and a published thesaurus (Macquarie Thesaurus). The algorithm determines estimates of the DPCs using the categories in the thesaurus as very coarse concepts and, notably, without requiring any sense-annotated data. Even though the use of only about 1000 concepts to represent the vocabulary of a language seems drastic, I show that the method achieves results better than the state-of-the-art in a number of natural language tasks. I show how cross-lingual DPCs can be created by combining text in one language with a thesaurus from another. Using these cross-lingual DPCs, we can solve problems in one, possibly resource-poor, language using a knowledge source from another, possibly resource-rich, language. I show that the approach is also useful in tasks that inherently involve two or more languages, such as machine translation and multilingual text summarization. The proposed approach is computationally inexpensive, it can estimate both semantic relatedness and semantic similarity, and it can be applied to all parts of speech. Extensive experiments on ranking word pairs as per semantic distance, real-word spelling correction, solving Reader's Digest word choice problems, determining word sense dominance, word sense disambiguation, and word translation show that the new approach is markedly superior to previous ones.

Semantic distance is a measure of how close or distant in meaning two units of language are. A large number of important natural language problems, including machine translation and word sense disambiguation, can be viewed as semantic distance problems. The two dominant approaches to estimating semantic distance are the WordNet-based semantic measures and the corpus-based distributional measures. In this thesis, I compare them, both qualitatively and quantitatively, and identify the limitations of each. This thesis argues that estimating semantic distance is essentially a property of concepts (rather than words) and that two concepts are semantically close if they occur in similar contexts. Instead of identifying the co-occurrence (distributional) profiles of words (distributional hypothesis), I argue that distributional profiles of concepts (DPCs) can be used to infer the semantic properties of concepts and indeed to estimate semantic distance more accurately. I propose a new hybrid approach to calculating semantic distance that combines corpus statistics and a published thesaurus (Macquarie Thesaurus). The algorithm determines estimates of the DPCs using the categories in the thesaurus as very coarse concepts and, notably, without requiring any sense-annotated data. Even though the use of only about 1000 concepts to represent the vocabulary of a language seems drastic, I show that the method achieves results better than the state-of-the-art in a number of natural language tasks. I show how cross-lingual DPCs can be created by combining text in one language with a thesaurus from another. Using these cross-lingual DPCs, we can solve problems in one, possibly resource-poor, language using a knowledge source from another, possibly resource-rich, language. I show that the approach is also useful in tasks that inherently involve two or more languages, such as machine translation and multilingual text summarization. The proposed approach is computationally inexpensive, it can estimate both semantic relatedness and semantic similarity, and it can be applied to all parts of speech. Extensive experiments on ranking word pairs as per semantic distance, real-word spelling correction, solving Reader's Digest word choice problems, determining word sense dominance, word sense disambiguation, and word translation show that the new approach is markedly superior to previous ones.

With the ever increase in biomedical literature, text-mining has emerged as an important technology to support bio-curation and search. Word sense disambiguation (WSD), the correct identification of terms in text in the light of ambiguity, is an important problem in text-mining. Since the late 1940s many approaches based on supervised (decision trees, naive Bayes, neural networks, support vector machines) and unsupervised machine learning (context-clustering, word-clustering, co-occurrence graphs) have been developed. Knowledge-based methods that make use of the WordNet computational lexicon have also been developed. But only few make use of ontologies, i.e. hierarchical controlled vocabularies, to solve the problem and none exploit inference over ontologies and the use of metadata from publications. This thesis addresses the WSD problem in biomedical ontologies by suggesting different approaches for word sense disambiguation that use ontologies and metadata. The "Closest Sense" method assumes that the ontology defines multiple senses of the term; it computes the shortest path of co-occurring terms in the document to one of these senses. The "Term Cooc" method defines a log-odds ratio for co-occurring terms including inferred co-occurrences. The "MetaData" approach trains a classifier on metadata; it does not require any ontology, but requires training data, which the other methods do not. These approaches are compared to each other when applied to a manually curated training corpus of 2600 documents for seven ambiguous terms from the Gene Ontology and MeSH. All approaches over all conditions achieve 80% success rate on average. The MetaData approach performs best with 96%, when trained on high-quality data. Its performance deteriorates as quality of the training data decreases. The Term Cooc approach performs better on Gene Ontology (92% success) than on MeSH (73% success) as MeSH is not a strict is-a/part-of, but rather a loose is-related-to hierarchy. The Closest Sense approach achieves on average 80% success rate. Furthermore, the thesis showcases applications ranging from ontology design to semantic search where WSD is important.