Spelling suggestions: "subject:"[een] TEXT MINING"" "subject:"[enn] TEXT MINING""
141 |
Organização flexível de documentos / Flexible organization of documentsRios, Tatiane Nogueira 25 March 2013 (has links)
Diversos métodos têm sido desenvolvidos para a organização da crescente quantidade de documentos textuais. Esses métodos frequentemente fazem uso de algoritmos de agrupamento para organizar documentos que referem-se a um mesmo assunto em um mesmo grupo, supondo que conteúdos de documentos de um mesmo grupo são similares. Porém, existe a possibilidade de que documentos pertencentes a grupos distintos também apresentem características semelhantes. Considerando esta situação, há a necessidade de desenvolver métodos que possibilitem a organização flexível de documentos, ou seja, métodos que possibilitem que documentos sejam organizados em diferentes grupos com diferentes graus de compatibilidade. O agrupamento fuzzy de documentos textuais apresenta-se como uma técnica adequada para este tipo de organização, uma vez que algoritmos de agrupamento fuzzy consideram que um mesmo documento pode ser compatível com mais de um grupo. Embora tem-se desenvolvido algoritmos de agrupamento fuzzy que possibilitam a organização flexível de documentos, tal organização é avaliada em termos do desempenho do agrupamento de documentos. No entanto, considerando que grupos de documentos devem possuir descritores que identifiquem adequadamente os tópicos representados pelos mesmos, de maneira geral os descritores de grupos tem sido extraídos utilizando alguma heurística sobre um conjunto pequeno de documentos, realizando assim, uma avaliação simples sobre o significado dos grupos extraídos. No entanto, uma apropriada extração e avaliação de descritores de grupos é importante porque os mesmos são termos representantes da coleção que identificam os tópicos abordados nos documentos. Portanto, em aplicações em que o agrupamento fuzzy é utilizado para a organização flexível de documentos, uma descrição apropriada dos grupos obtidos é tão importante quanto um bom agrupamento, uma vez que, neste tipo de agrupamento, um mesmo descritor pode indicar o conteúdo de mais de um grupo. Essa necessidade motivou esta tese, cujo objetivo foi investigar e desenvolver métodos para a extração de descritores de grupos fuzzy para a organização flexível de documentos. Para cumprir esse objetivo desenvolveu se: i) o método SoftO-FDCL (Soft Organization - Fuzzy Description Comes Last ), pelo qual descritores de grupos fuzzy at são extraídos após o processo de agrupamento fuzzy, visando identicar tópicos da organização flexível de documentos independentemente do algoritmo de agrupamento fuzzy utilizado; ii) o método SoftO-wFDCL ( Soft Organization - weighted Fuzzy Description Comes Last ), pelo qual descritores de grupos fuzzy at também são extraídos após o processo de agrupamento fuzzy utilizando o grau de pertinência dos documentos em cada grupo, obtidos do agrupamento fuzzy, como fator de ponderação dos termos candidatos a descritores; iii) o método HSoftO-FDCL (Hierarchical Soft Organization - Fuzzy Description Comes Last ), pelo qual descritores de grupos fuzzy hierárquicos são extraídos após o processo de agrupamento hierárquico fuzzy, identificando tópicos da organização hierárquica flexível de documentos. Adicionalmente, apresenta-se nesta tese uma aplicação do método SoftO-FDCL no contexto do programa de educação médica continuada canadense, reforçando a utilidade e aplicabilidade da organização flexível de documentos / Several methods have been developed to organize the growing number of textual documents. Such methods frequently use clustering algorithms to organize documents with similar topics into clusters. However, there are situations when documents of dffierent clusters can also have similar characteristics. In order to overcome this drawback, it is necessary to develop methods that permit a soft document organization, i.e., clustering documents into different clusters according to different compatibility degrees. Among the techniques that we can use to develop methods in this sense, we highlight fuzzy clustering algorithms (FCA). By using FCA, one of the most important steps is the evaluation of the yield organization, which is performed considering that all analyzed topics are adequately identified by cluster descriptors. In general, cluster descriptors are extracted using some heuristic over a small number of documents. The adequate extraction and evaluation of cluster descriptors is important because they are terms that represent the collection and identify the topics of the documents. Therefore, an adequate description of the obtained clusters is as important as a good clustering, since the same descriptor might identify one or more clusters. Hence, the development of methods to extract descriptors from fuzzy clusters obtained for soft organization of documents motivated this thesis. Aiming at investigating such methods, we developed: i) the SoftO-FDCL (Soft Organization - Fuzzy Description Comes Last) method, in which descriptors of fuzzy clusters are extracted after clustering documents, identifying topics regardless the adopted fuzzy clustering algorithm; ii) the SoftO-wFDCL (Soft Organization - weighted Fuzzy Description Comes Last) method, in which descriptors of fuzzy clusters are also extracted after the fuzzy clustering process using the membership degrees of the documents as a weighted factor for the candidate descriptors; iii) the HSoftO-FDCL (Hierarchical Soft Organization - Fuzzy Description Comes Last) method, in which descriptors of hierarchical fuzzy clusters are extracted after the hierarchical fuzzy clustering process, identifying topics by means of a soft hierarchical organization of documents. Besides presenting these new methods, this thesis also discusses the application of the SoftO-FDCL method on documents produced by the Canadian continuing medical education program, presenting the utility and applicability of the soft organization of documents in real-world scenario
|
142 |
Uso do minerador de textos sobek como ferramenta de apoio à compreensão textualEpstein, Daniel January 2017 (has links)
A presente tese tem por objetivo investigar os efeitos do uso do minerador de textos Sobek no processo de leitura e compreensão textual de estudantes. Este minerador de textos é capaz de extrair informações relevantes de textos e representá-las de forma gráfica. Esta tese está apoiada nas teorias de aprendizagem significativa, de uso de mapas conceituais para representação de conhecimento e em pesquisas que apontam que representações gráficas de palavras auxiliam na leitura de textos e na sua decodificação. De acordo com a pesquisa de David Ausubel, a aprendizagem significativa ocorre através da assimilação de novos conceitos e ideias e associação destas ao conhecimento que a pessoa já possui. Através da utilização de um minerador de textos com representação gráfica de informações, busca-se apresentar aos estudantes uma representação visual de textos. Esta representação se assemelha a de um mapa conceitual, de forma a auxiliar no processo de compreensão e assimilação de informações pelos estudantes. Nesta representação, ligações entre termos considerados relevantes pelo minerador auxiliam no entendimento destes termos e simbolizam relações presentes no texto, fato esse que pode auxiliar os estudantes a compreenderem melhor o texto e relacionarem novas informações àquelas que já possuem Nesta tese, foi realizado um estudo para auxiliar estudantes nas atividades relacionadas ao letramento. A pesquisa se caracteriza como mista (qualitativa e quantitativa). A coleta de dados se deu a partir da aplicação de questionários com professores e alunos, além de avaliações com o objetivo de verificar contribuições do uso da ferramenta a partir de seu uso do ponto de vista do letramento. Como resultado, encontramos que estudantes que utilizaram o Sobek obtiveram um número mais elevado de respostas corretas nas atividades de interpretação de textos. Em média, os alunos acertaram 66% das questões quando utilizando o minerador de textos Sobek, contra apenas 47% das questões que eram respondidas sem o apoio do minerador. Outro resultado apresentado é o alto grau de satisfação de alunos e professores quanto à tecnologia e seu uso em sala de aula. Além destes resultados, obtivemos uma avaliação acerca da capacidade do minerador de textos de extrair termos considerados relevantes ao texto. / This thesis aimed to investigate the effects of using Sobek Text Miner to improve literacy. Sobek is a tool capable of extracting relevant information from texts and representing them in a graphical way. The thesis is supported by meaningful learning theory, conceptual maps theory and several research theories which indicate that graphical representation of words may improve reading capability and word decoding. According to David Ausubel, meaningful learning occurs through the assimilation of new concepts and ideas and association of those to what the person already knows. Using text mining with graphical representation of information, we seek to provide students with a graphical representation of a text. This text representation is similar to a concept map, helping students assimilate and comprehend that information. In Sobek’s representation, the relationship between terms considered relevant to text comprehension may assist students to better understand the meaning of each term and demonstrate relationships that are presented in the text, improving context comprehension. Furthermore, the relationship between terms may help information assimilation, once it relates the new information with previous known information This project conducted a study using Sobek text miner in classroom to support student’s literacy. In order to assess the tool’s possible benefits in reading and comprehension activities, we designed a series of classroom activities. To evaluate those activities, qualitative and quantitative approaches were used. The study was conducted in two primary schools, with students from 5th grade and 8th grade. Interviews were also made with the teachers and students, inquiring them about the tool's and main functions and its ability to help students from a literacy point of view. The study shows that students answered more correct question when using Sobek than when no support technology was used. Also, both students and teachers approved the software and agreed that it does improve student’s text comprehension. It also describes an evaluation of Sobek's capability to extract terms considered relevant for text comprehension.
|
143 |
Mineração de textos aplicada na previsão e detecção de eventos adversos no Hospital de Clínicas de Porto AlegreSilva, Daniel Antonio da January 2017 (has links)
Este trabalho apresenta os resultados de uma pesquisa que teve como objetivo avaliar o desempenho de métodos de mineração de textos na previsão e detecção de Eventos Adversos (EA). A primeira etapa foi a revisão sistemática da literatura que buscou identificar os métodos de mineração de textos e as áreas da saúde que esses estão sendo aplicados para prever e detectar EA. Após essa etapa foi realizada uma aplicação de métodos de mineração de textos para prever Infecções do Sítio Cirúrgico (ISC) a partir do texto livre de descrições cirúrgicas no Hospital de Clínicas de Porto Alegre (HCPA). Por fim, métodos de mineração de textos foram aplicados para detectar ISC a partir do texto das evoluções de pacientes 30 (trinta) dias após uma cirurgia. Como resultados, destaca-se a identificação dos melhores métodos de pré-processamento e mineração de textos para prever e detectar ISC no HCPA, podendo ser aplicados a outros EA. O método Stochastic Gradient Descent (SGD) apresentou o melhor desempenho, 79,7% de ROC-AUC na previsão de EA. Já para detecção de EA o melhor método foi o Logistic Regression, com desempenho 80,6% de ROC-AUC. Os métodos de mineração de textos podem ser usados para apoiar de maneira eficaz a previsão e detecção de EA, direcionando ações de vigilância para a melhoria da segurança do paciente. / This work presents the results of a research that aimed to evaluate the performance of text mining methods in the prediction and detection of Adverse Events (AE). The first step was the systematic review of the literature that sought to identify the methods of text mining and the health areas they are being applied to predict and detect AE. After this step, an application of text mining methods was performed to predict Surgical Site Infections (SSI) from the free text of medical records at Hospital de Clínicas de Porto Alegre (HCPA). Finally, text mining methods were applied to detect SSI from the text of medical records 30 (thirty) days after surgery. As results, is highlight the identification of the best methods of pre-processing and text mining to predict and detect SSI in the HCPA, and can be applied to other AE. The Stochastic Gradient Descent (SGD) presented the best performance, 79.7% of ROC-AUC in the prediction of AE. Already for the detection of AE the best method was the Logistic Regression, with performance 80.6% of ROC-AUC. Text mining methods can be used to effectively support the prediction and detection of AE by directing surveillance actions to improve patient safety.
|
144 |
An empirical study on object-oriented software dependencies : logical, structural and semanticAjienka, Nemitari Miebaka January 2018 (has links)
Three of the most widely studied software dependency types are the structural, logical and semantic dependencies. Logical dependencies capture the degree of co-change between software artifacts. Semantic dependencies capture the degree to which artifacts, comments and names are related. Structural dependencies capture the dependencies in the source code of artifacts. Prior studies show that a combination of dependency analysis (e.g., semantic and logical analysis) improves accuracy when predicting which artifacts are likely to be impacted by ripple effects of software changes (though not to a large extent) compared to individual approaches. In addition, some dependencies could be hidden dependencies when an analysis of one dependency type (e.g., logical) does not reveal artifacts only linked by another dependency type (semantic). While previous studies have focused on combining dependency information with minimal benefits, this Thesis explores the consistency of these measurements, and whether hidden dependencies arise between artifacts, and in any of the axes studied. In this Thesis, 79 Java projects are empirically studied to investigate (i) the direct influence and the degree of overlap between dependency types on three axes (logical - structural (LSt); logical - semantic (LSe); structural - semantic (StSe)) (structural, logical and semantic), and (ii) the presence of hidden coupling on the axes. The results show that a high proportion of hidden dependencies can be detected on the LSt and StSe axes. Notwithstanding, the LSe axis shows a much smaller proportion of hidden dependencies. Practicable refactoring methods to mitigate hidden dependencies are proposed in the Thesis and discussed with examples.
|
145 |
Sistema de recomendação de objeto de aprendizagem baseado em postagens extraídas do ambiente virtual de aprendizagemSilva, Reinaldo de Jesus da January 2016 (has links)
Os fóruns de discussões apresentam-se com umas das ferramentas de interação utilizadas nos ambientes virtuais de aprendizagem (AVAs). Esta pesquisa tem como objetivo propor um sistema computacional para recomendação de Objeto de Aprendizagem (OA), levando em consideração as postagens feitas de dentro dos fóruns de um Ambiente Virtual de Aprendizagem (AVA). A metodologia utilizada foi a pesquisa qualitativa, dos tipos descritiva e explicativa. Esse sistema identifica as palavras-chave nos fóruns de um AVA; usam as palavras-chave como indícios dos interesses dos usuários; classifica (atributos pesos) as palavras mais relevantes (Hot Topics); submete a um mecanismo de busca (repositório), neste trabalho foram usados os motores de busca, para fins de teste e oferece os resultados da busca aos usuários. As contribuições deste sistema para os sujeitos participantes desta pesquisa são: recomendação automática de OA para os alunos e professores; aplicação de mineração de dados para sistema gestão educacional; técnica de mineração de textos, utilizando algoritmo TF*PDF (Term Frequency * Proportional Document Frequency) e integração do AVA com repositório digital. Para validar o sistema de recomendação de OA em um AVA foi desenvolvido protótipo do sistema com uma amostra, contendo vinte e cinco alunos e cinco professores de duas turmas das disciplinas de Modelagem de Banco de Dados e Interface de Usuários e Sistemas Computacionais do curso de Engenharia de Computação da Universidade Estadual do Maranhão. O estudo realizado sobre o tema, e relatado nessa tese, tem como foco a recomendação de OA nos fóruns de um AVA. A avaliação e validação realizadas, através de protótipo do sistema com professores e alunos evidenciaram que o sistema de recomendação de Web Services (RECOAWS) proposto atende às expectativas e pode apoiar professores e alunos, nas suas atividades pedagógicas, dentro dos fóruns. / Discussion forums get present with one of interaction tools used in virtual learning environments (VLEs). This research aims to propose a computational system for Learning Object recommendation (LO), taking into account the posts made from within the forums of a Virtual Learning Environment (VLE). The methodology used was a qualitative study of descriptive and explanatory types. This system identifies the keywords in the forums of a VLE; It uses the keywords as evidence of the interests of users; ranks (attributes weights) the most relevant words (Hot Topics); It submits to a search engine (repository), this work were used search engines for testing purposes and provides the search results to users. The contributions of this system to the participants in this study are: automatic recommendation of LO for students and teachers; data mining application to educational management system; text mining techniques, using TF * PDF algorithm (Term Frequency * Proportional Document Frequency) and integration of the VLE with digital repository. To validate the LO recommendation system in a VLE was developed prototype system with a sample, with twenty-five students and five teachers from two classes of database modeling disciplines and User Interface and Computational Systems of Engineering course Computing of the State University of Maranhão. The study on the subject, and reported in this thesis is focused on LO recommendation in the forums of a VLE. The evaluation and validation performed by the prototype system with teachers and students showed that the Web Services recommendation system (RecoaWS) proposed meets expectations and can support teachers and students in their educational activities within the forums.
|
146 |
Aprendizado não supervisionado de hierarquias de tópicos a partir de coleções textuais dinâmicas / Unsupervised learning of topic hierarchies from dynamic text collectionsRicardo Marcondes Marcacini 19 May 2011 (has links)
A necessidade de extrair conhecimento útil e inovador de grandes massas de dados textuais, tem motivado cada vez mais a investigação de métodos para Mineração de Textos. Dentre os métodos existentes, destacam-se as iniciativas para organização de conhecimento por meio de hierarquias de tópicos, nas quais o conhecimento implícito nos textos é representado em tópicos e subtópicos, e cada tópico contém documentos relacionados a um mesmo tema. As hierarquias de tópicos desempenham um papel importante na recupera ção de informação, principalmente em tarefas de busca exploratória, pois permitem a análise do conhecimento de interesse em diversos níveis de granularidade e exploração interativa de grandes coleções de documentos. Para apoiar a construção de hierarquias de tópicos, métodos de agrupamento hierárquico têm sido utilizados, uma vez que organizam coleções textuais em grupos e subgrupos, de forma não supervisionada, por meio das similaridades entre os documentos. No entanto, a maioria dos métodos de agrupamento hierárquico não é adequada em cenários que envolvem coleções textuais dinâmicas, pois são exigidas frequentes atualizações dos agrupamentos. Métodos de agrupamento que respeitam os requisitos existentes em cenários dinâmicos devem processar novos documentos assim que são adicionados na coleção, realizando o agrupamento de forma incremental. Assim, neste trabalho é explorado o uso de métodos de agrupamento incremental para o aprendizado não supervisionado de hierarquias de tópicos em coleções textuais dinâmicas. O agrupamento incremental é aplicado na construção e atualização de uma representação condensada dos textos, que mantém um sumário das principais características dos dados. Os algoritmos de agrupamento hierárquico podem, então, ser aplicados sobre as representa ções condensadas, obtendo-se a organização da coleção textual de forma mais eficiente. Foram avaliadas experimentalmente três estratégias de agrupamento incremental da literatura, e proposta uma estratégia alternativa mais apropriada para hierarquias de tópicos. Os resultados indicaram que as hierarquias de tópicos construídas com uso de agrupamento incremental possuem qualidade próxima às hierarquias de tópicos construídas por métodos não incrementais, com significativa redução do custo computacional / The need to extract new and useful knowledge from large textual collections has motivated researchs on Text Mining methods. Among the existing methods, initiatives for the knowledge organization by topic hierarchies are very popular. In the topic hierarchies, the knowledge is represented by topics and subtopics, and each topic contains documents of similar content. They play an important role in information retrieval, especially in exploratory search tasks, allowing the analysis of knowledge in various levels of granularity and interactive exploration of large document collections. Hierarchical clustering methods have been used to support the construction of topic hierarchies. These methods organize textual collections in clusters and subclusters, in an unsupervised manner, using similarities among documents. However, most existing hierarchical clustering methods is not suitable for scenarios with dynamic text collections, since frequent clustering updates are necessary. Clustering methods that meet these requirements must process new documents that are inserted into textual colections, in general, through incremental clustering. Thus, we studied the incremental clustering methods for unsupervised learning of topic hierarchies for dynamic text collections. The incremental clustering is used to build and update a condensed representation of texts, which maintains a summary of the main features of the data. The hierarchical clustering algorithms are applied in these condensed representations, obtaining the textual organization more efficiently. We experimentally evaluate three incremental clustering algorithms available in the literature. Also, we propose an alternative strategy more appropriate for construction of topic hieararchies. The results indicated that the topic hierarchies construction using incremental clustering have quality similar to non-incremental methods. Furthermore, the computational cost is considerably reduced using incremental clustering methods
|
147 |
Mineração textual e produção de fanfictions : processos desencadeadores de oportunidades de letramento no ensino de língua estrangeiraBarcellos, Patrícia da Silva Campelo Costa January 2013 (has links)
Esta tese tem por objetivo investigar como o letramento em língua estrangeira (LE) pode ser apoiado pelo uso de um recurso digital passível de auxiliar os processos de leitura e produção textual. Assim, a presente pesquisa baseia-se nos estudos de Feldman e Sanger (2006) acerca da mineração de textos e nas pesquisas de Black (2007, 2009) sobre a incorporação de um gênero textual característico da internet (fanfiction) na aprendizagem de línguas. Através da utilização de um recurso de mineração de texto (Sobek), a partir do qual ocorre a extração dos termos mais recorrentes em um texto, os participantes deste estudo criaram narrativas, em meio digital. Os doze alunos participantes da pesquisa utilizaram a ferramenta Sobek como mediadora da produção de histórias conhecidas como fanfictions, nas quais novas tramas são criadas a partir de elementos culturais já reconhecidos na mídia. Os informantes eram seis graduandos em Letras e seis alunos de um curso de extensão, ambos os grupos na Universidade Federal do Rio Grande do Sul (UFRGS). Na tarefa proposta, cada aprendiz leu uma fanfiction de sua escolha, publicada na web, e utilizou a ferramenta de mineração para formar grafos com os termos mais recorrentes da história. Durante tal processo, o aluno tinha oportunidade de fazer associações entre as expressões do texto, de modo a formar, na ferramenta Sobek, uma imagem em rede (grafo) que representasse termos recorrentes nesse gênero textual (tais como o uso de tempos verbais no passado e adjetivos para caracterizar personagens e contexto). Posteriormente, esse grafo foi repassado a um colega, que assim iniciou seu processo de composição com base nessa imagem representativa do texto. A partir da análise dos dados, observou-se que a utilização da ferramenta digital deu suporte à produção textual em LE, e sua subsequente prática de letramento, visto que os autores se apoiaram no recurso de mineração para criar suas narrativas fanfiction. / This doctoral thesis aims at investigating how literacy in a foreign language (FL) may be supported by the use of a digital resource which can help the processes of reading and writing. Thus, the present research is based on studies by Feldman and Sanger (2006) about text mining, and on research by Black (2007, 2009) about the incorporation of a textual genre characteristic of the Internet (fanfiction) in language learning. Through the use of a text mining resource (Sobek), which promotes the extraction of frequent terms present in a text, the participants of this study created narratives, in digital media. The twelve students who participated in the research used the tool Sobek to mediate the production of stories known as fanfictions, in which new plots are created from cultural elements already recognized in the media. The participants were six undergraduate students of Languages and six students who were part of an extension course, both groups at the Federal University of Rio Grande do Sul (UFRGS). In the proposed task, each student read a fanfiction of his/her choice, which was published on a website, and used the mining tool to develop graphs with the recurrent terms found in the story. During this process, the student had the opportunity to make associations between expressions from the text, using the software Sobek, so as to form an image (graph) that represented terms used in this textual genre (such as the use of verbal tenses in the past and adjectives to describe characters and context). Later, this graph was forwarded to a peer, who then began his/her writing process based on this picture originated from a text. From the data analysis, it was observed that the use of a digital tool supported the text production in the FL, and its following practice of literacy, as the authors relied on the mining resource to create their fanfictions.
|
148 |
Systematic Analysis of the Factors Contributing to the Variation and Change of the MicrobiomeJanuary 2018 (has links)
abstract: Understanding changes and trends in biomedical knowledge is crucial for individuals, groups, and institutions as biomedicine improves people’s lives, supports national economies, and facilitates innovation. However, as knowledge changes what evidence illustrates knowledge changes? In the case of microbiome, a multi-dimensional concept from biomedicine, there are significant increases in publications, citations, funding, collaborations, and other explanatory variables or contextual factors. What is observed in the microbiome, or any historical evolution of a scientific field or scientific knowledge, is that these changes are related to changes in knowledge, but what is not understood is how to measure and track changes in knowledge. This investigation highlights how contextual factors from the language and social context of the microbiome are related to changes in the usage, meaning, and scientific knowledge on the microbiome. Two interconnected studies integrating qualitative and quantitative evidence examine the variation and change of the microbiome evidence are presented. First, the concepts microbiome, metagenome, and metabolome are compared to determine the boundaries of the microbiome concept in relation to other concepts where the conceptual boundaries have been cited as overlapping. A collection of publications for each concept or corpus is presented, with a focus on how to create, collect, curate, and analyze large data collections. This study concludes with suggestions on how to analyze biomedical concepts using a hybrid approach that combines results from the larger language context and individual words. Second, the results of a systematic review that describes the variation and change of microbiome research, funding, and knowledge are examined. A corpus of approximately 28,000 articles on the microbiome are characterized, and a spectrum of microbiome interpretations are suggested based on differences related to context. The collective results suggest the microbiome is a separate concept from the metagenome and metabolome, and the variation and change to the microbiome concept was influenced by contextual factors. These results provide insight into how concepts with extensive resources behave within biomedicine and suggest the microbiome is possibly representative of conceptual change or a preview of new dynamics within science that are expected in the future. / Dissertation/Thesis / Doctoral Dissertation Biology 2018
|
149 |
Modeling words for online sexual behavior surveillance and clinical text information extractionFries, Jason Alan 01 July 2015 (has links)
How do we model the meaning of words? In domains like information retrieval, words have classically been modeled as discrete entities using 1-of-n encoding, a representation that elides most of a word's syntactic and semantic structure. Recent research, however, has begun exploring more robust representations called word embeddings. Embeddings model words as a parameterized function mapping into an n-dimensional continuous space and implicitly encode a number of interesting semantic and syntactic properties. This dissertation examines two application areas where existing, state-of-the-art terminology modeling improves the task of information extraction (IE) -- the process of transforming unstructured data into structured form. We show that a large amount of word meaning can be learned directly from very large document collections.
First, we explore the feasibility of mining sexual health behavior data directly from the unstructured text of online “hookup" requests. The Internet has fundamentally changed how individuals locate sexual partners. The rise of dating websites, location-aware smartphone apps like Grindr and Tinder that facilitate casual sexual encounters (“hookups"), as well as changing trends in sexual health practices all speak to the shifting cultural dynamics surrounding sex in the digital age. These shifts also coincide with an increase in the incidence rate of sexually transmitted infections (STIs) in subpopulations such as young adults, racial and ethnic minorities, and men who have sex with men (MSM). The reasons for these increases and their possible connections to Internet cultural dynamics are not completely understood. What is apparent, however, is that sexual encounters negotiated online complicate many traditional public health intervention strategies such as contact tracing and partner notification. These circumstances underline the need to examine online sexual communities using computational tools and techniques -- as is done with other social networks -- to provide new insight and direction for public health surveillance and intervention programs.
One of the central challenges in this task is constructing lexical resources that reflect how people actually discuss and negotiate sex online. Using a 2.5-year collection of over 130 million Craigslist ads (a large venue for MSM casual sexual encounters), we discuss computational methods for automatically learning terminology characterizing risk behaviors in the MSM community. These approaches range from keyword-based dictionaries and topic modeling to semi-supervised methods using word embeddings for query expansion and sequence labeling. These methods allow us to gather information similar (in part) to the types of questions asked in public health risk assessment surveys, but automatically aggregated directly from communities of interest, in near real-time, and at geographically high-resolution. We then address the methodological limitations of this work, as well as the fundamental validation challenges posed by the lack of large-scale sexual sexual behavior survey data and limited availability of STI surveillance data.
Finally, leveraging work on terminology modeling in Craigslist, we present new research exploring representation learning using 7 years of University of Iowa Hospitals and Clinics (UIHC) clinical notes. Using medication names as an example, we show that modeling a low-dimensional representation of a medication's neighboring words, i.e., a word embedding, encodes a large amount of non-obvious semantic information. Embeddings, for example, implicitly capture a large degree of the hierarchical structure of drug families as well as encode relational attributes of words, such as generic and brand names of medications. These representations -- learned in a completely unsupervised fashion -- can then be used as features in other machine learning tasks. We show that incorporating clinical word embeddings in a benchmark classification task of medication labeling leads to a 5.4% increase in F1-score over a baseline of random initialization and a 1.9% over just using non-UIHC training data. This research suggests clinical word embeddings could be shared for use in other institutions and other IE tasks.
|
150 |
Computational methods for mining health communications in web 2.0Bhattacharya, Sanmitra 01 May 2014 (has links)
Data from social media platforms are being actively mined for trends and patterns of interests. Problems such as sentiment analysis and prediction of election outcomes have become tremendously popular due to the unprecedented availability of social interactivity data of different types. In this thesis we address two problems that have been relatively unexplored. The first problem relates to mining beliefs, in particular health beliefs, and their surveillance using social media. The second problem relates to investigation of factors associated with engagement of U.S. Federal Health Agencies via Twitter and Facebook.
In addressing the first problem we propose a novel computational framework for belief surveillance. This framework can be used for 1) surveillance of any given belief in the form of a probe, and 2) automatically harvesting health-related probes. We present our estimates of support, opposition and doubt for these probes some of which represent true information, in the sense that they are supported by scientific evidence, others represent false information and the remaining represent debatable propositions. We show for example that the levels of support in false and debatable probes are surprisingly high. We also study the scientific novelty of these probes and find that some of the harvested probes with sparse scientific evidence may indicate novel hypothesis. We also show the suitability of off-the-shelf classifiers for belief surveillance. We find these classifiers are quite generalizable and can be used for classifying newly harvested probes. Finally, we show the ability of harvesting and tracking probes over time. Although our work is focused in health care, the approach is broadly applicable to other domains as well.
For the second problem, our specific goals are to study factors associated with the amount and duration of engagement of organizations. We use negative binomial hurdle regression models and Cox proportional hazards survival models for these. For Twitter, the hurdle analysis shows that presence of user-mention is positively associated with the amount of engagement while negative sentiment has inverse association. Content of tweets is also equally important for engagement. The survival analyses indicate that engagement duration is positively associated with follower count. For Facebook, both hurdle and survival analyses show that number of page likes and positive sentiment are correlated with higher and prolonged engagement while few content types are negatively correlated with engagement. We also find patterns of engagement that are consistent across Twitter and Facebook.
|
Page generated in 0.0944 seconds