In recent years, geography has acquired a great importance in the context of Information Retrieval (IR) and, in general, of the automated processing of information in text. Mobile devices that are able to surf the web and at the same time inform about their position are now a common reality, together with applications that can exploit this data to provide users with locally customised information, such as directions or advertisements. Therefore, it is important to deal properly with the geographic information that is included in electronic texts. The majority of such kind of information is contained as place names, or toponyms. Toponym ambiguity represents an important issue in Geographical Information Retrieval (GIR), due to the fact that queries are geographically constrained. There has been a struggle to nd speci c geographical IR methods that actually outperform traditional IR techniques. Toponym ambiguity may constitute a relevant factor in the inability of current GIR systems to take advantage from geographical knowledge. Recently, some Ph.D. theses have dealt with Toponym Disambiguation (TD) from di erent perspectives, from the development of resources for the evaluation of Toponym Disambiguation (Leidner (2007)) to the use of TD to improve geographical scope resolution (Andogah (2010)). The Ph.D. thesis presented here introduces a TD method based on WordNet and carries out a detailed study of the relationship of Toponym Disambiguation to some IR applications, such as GIR, Question Answering (QA) and Web retrieval. The work presented in this thesis starts with an introduction to the applications in which TD may result useful, together with an analysis of the ambiguity of toponyms in news collections. It could not be possible to study the ambiguity of toponyms without studying the resources that are used as placename repositories; these resources are the equivalent to language dictionaries, which provide the di erent meanings of a given word. / Buscaldi, D. (2010). Toponym Disambiguation in Information Retrieval [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/8912

Made available in DSpace on 2016-03-22T17:26:21Z (GMT). No. of bitstreams: 1 dissertacao_mestrado_cleber.pdf: 565462 bytes, checksum: 906465b0884050d40a2c09bf52b60526 (MD5) Previous issue date: 2009-03-23 / Georeferencing of texts, that is, the identification of the geographical context of texts is becoming popular in the Web due to the high demand for geographical information and due to the raising of services for query and retrieval like Google Earth (geobrowsers). The main challenge is to relate texts to geographical locations. These associations are stored in structures called gazetteers. Although there are gazetteers like Geonames and TGN, they fail in coverage, lacking information about some countries, and they also fail by weak specialization, lacking detailed references to locations (fine granularity) as for example names of streets, squares, monuments, rivers, neighborhoods, etc. This kind of information that acts as indirect references to geographical locations is defined as Location Indicators . This dissertation presents an approach that identifies Location Indicators related to geographical locations, by analyzing texts of news published in the Web. The goal is to enrich create gazetteers with the identified relations and then perform geo-referencing of news. Location Indicators include non-geographical entities that are dynamic and may change along the time. The use of news published in the Web is a useful way to discover Location Indicators, covering a great number of locations and maintaining detailed information about each location. Different training news corpora are compared for the creation of gazetteers and evaluated by their ability to correctly identify cities in texts of news Georeferencing of texts, that is, the identification of the geographical context of texts is becoming popular in the Web due to the high demand for geographical information and due to the raising of services for query and retrieval like Google Earth (geobrowsers). The main challenge is to relate texts to geographical locations. These associations are stored in structures called gazetteers. Although there are gazetteers like Geonames and TGN, they fail in coverage, lacking information about some countries, and they also fail by weak specialization, lacking detailed references to locations (fine granularity) as for example names of streets, squares, monuments, rivers, neighborhoods, etc. This kind of information that acts as indirect references to geographical locations is defined as Location Indicators . This dissertation presents an approach that identifies Location Indicators related to geographical locations, by analyzing texts of news published in the Web. The goal is to enrich create gazetteers with the identified relations and then perform geo-referencing of news. Location Indicators include non-geographical entities that are dynamic and may change along the time. The use of news published in the Web is a useful way to discover Location Indicators, covering a great number of locations and maintaining detailed information about each location. Different training news corpora are compared for the creation of gazetteers and evaluated by their ability to correctly identify cities in texts of news / Com o advento da Internet e o crescente número de informações disponíveis torna-se necessária a definição de estratégias especiais que permitam aos usuários o acesso rápido a informações relevantes. Como a Web possui grande volume de informações principalmente com o foco geográfico torna-se necessário recuperar e estruturar essas informações de forma a poder relacioná-las com o contexto e realidade das pessoas através de métodos e sistemas automáticos. Para isso uma das necessidades é possibilitar o georreferenciamento dos textos, ou seja, identificar as entidades geográficas presentes e associá-las com sua correta localização espacial. Nesse sentido, os topônimos (ex: nomes de localidades como cidades, países, etc.), devido à possibilidade de identificar de forma precisa determinada região espacial, apresentam-se como ideais para a identificação do contexto geográfico dos textos. Essa tarefa, denominada de Resolução de Topônimos apresenta, no entanto, desafios importantes principalmente do ponto de vista lingüístico, já que uma localidade pode possuir variados tipos de ambigüidade. Com relação a isso a principal estratégia para superar estes problemas compreende a identificação de evidências que auxiliem na identificação e desambiguação das localidades nos textos. Para essa verificação são utilizados geralmente os serviços de um ou mais dicionários toponímicos (Gazetteers). Como são criados de forma manual eles apresentam, no entanto deficiência de informações relacionadas principalmente a entidades que podem identificar, embora de forma indireta, determinados tipos de lugares como ruas, praças, universidades etc., as quais são definidas como Indicadores de Localidade. O presente trabalho propõe uma abordagem para a recuperação dessas entidades aproveitando para isso o caráter geográfico das informações jornalísticas. Para ilustrar a viabilidade do processo diferentes tipos de corpora de notícias foram testados e comparados pela habilidade de criação de Gazetteers com os Indicadores recuperados, sendo os Gazetteers avaliados então pela capacidade de identificação das cidades relacionadas às notícias testadas. Os resultados demonstram a utilidade da abordagem para o enriquecimento de Gazetteers e consequentemente para a recuperação de Indicadores de Localidade com maior simplicidade e extensibilidade que os trabalhos atuais

隨著GIS的發展，地圖定位成為空間查詢中極為普遍的行為。一般地圖定位大多透過地址來進行，但是在缺乏地址的情況之下，進行地圖上之定位變成極為困難之事。 本論文嘗試對手繪地圖在真實地圖上進行定位，我們提出了一套機制，使用者可以隨意地以手繪方式繪製街道圖與地標，透過我們提出的方法，即可自動地在真實的地圖上進行定位。 論文中，我們使用相鄰街廓中之地標配置與相鄰之交叉路口之地標配置等變數組成的表示法來描述地圖。我們將手繪地圖與真實地圖轉換成這些表示法，並透過字串編輯距離、圖同構等關係來比較手繪地圖與真實地圖之相似度，從而對手繪地圖進行定位。 實作中，我們挑選了幾處真實場景在台北市地圖中進行比對並觀察其結果。系統採用之地標包括政府機構(如派出所、消防隊、區公所等)、學校、醫院等資料。在實驗中，應用這套表示法可成功的定位出使用者所輸入之各場景所在位置。另外，透過控制相似度門檻值，我們可以調整辨識之精確度，不至於錯失可能之定位結果。 / As the widely spread of the GIS applications, map localization becomes one of the most important features in the spatial information retrieval. Normally, map localization is done through street addresses. Without this information, map localization becomes very difficult. In this research, we are trying to do map localization using hand drawing maps. We proposed a mechanism that can localize the user's drawing map in the reference map automatically. Our approaches use the landmark configurations of the adjacent street blocks as well as the landmark configurations of the adjacent street intersections as the descriptors in representing a map. The user's hand drawn maps and the reference maps are converted into these representations. The string editing distances and graph isomorphism are used in determining the similarities between the hand drawn map and the reference map. The map localization can be done by comparing these similarities. We used various real scenes in Taipei City to verify our systems. The landmarks we used including police offices, fire stations, county offices, schools and hospitals, etc. The experimental results shown that our system can localize the user's input successfully. Moreover, by controlling thresholds in similarity analysis, we can adjust the system's accuracy that reduces possibility of miss localizations.