| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊工程學系 | zh_TW |
| DC.creator | 李淑瑩 | zh_TW |
| DC.creator | Shu-Ying Li | en_US |
| dc.date.accessioned | 2009-7-28T07:39:07Z | |
| dc.date.available | 2009-7-28T07:39:07Z | |
| dc.date.issued | 2009 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:444/thesis/view_etd.asp?URN=965202104 | |
| dc.contributor.department | 資訊工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 地址資訊和人們的日常生活息息相關，人們常需要透過網路查詢相關實體商店、學校或組織的地址，再經由地圖標示服務確定其實際方位。然而並不是每一個網站同時提供地址與地圖標示的功能，因此本研究目的是希望設計一個能從網頁中自動擷取英文地址的服務，並結合地圖標示功能，將擷取到的地址以及其相關資訊，一併標示在地圖上，提供使用者簡單方便的地圖標記資訊服務。我們的系統分為兩個部分，第一部分，將網頁透過條件式隨機域的方式訓練出地址擷取的模型，輸入的網頁經過此模型的測試過程後並擷取地址；第二部份，則以擷取到的地址為基礎，在網頁中擷取與地址相關的資訊，找出包含地址和相關資訊的地址區塊邊界，並且針對包含多餘資訊的區塊提出調整的作法。實驗結果得知，我們的地址擷取效能可以提升F-measure至0.913，同時對於八成六的資料可以正確的擷取到相關資訊。
| zh_TW |
| dc.description.abstract | Address Information is closely linked to people’’s daily life. People often need to query addresses of related brick-and-mortar shopping malls、schools and organization. And using the service of map marking identified the real direction. But there are not all web pages providing addresses and facility of map marking. Therefore, designing a service of extracting English addresses automatically from web pages is the goal of our research. And the service combines the facility of map marking and marks the extracted addresses and the related information on the map. The service provides users in a convenient and easy way to using the information service of map marking. Our system is divided into two steps: the first step is using Conditional Random fields to train the model of address extraction. Page we input enters the testing process of model of address extraction and outputs the segment of address. The second step is using extracted addresses as landmarks to extract related information and finding out the correct boundary of address blocks. In terms of the result of experiment, the F-measure of extraction by Conditional Random field is up to 0.913. And we also propose the method of adjustment to revise the incorrect boundary. The accuracy after adjusting is from 0.8506 to 0.8689.
| en_US |
| DC.subject | Conditional Random Fields | zh_TW |
| DC.subject | Postal Addresses | zh_TW |
| DC.subject | 地址擷取 | en_US |
| DC.subject | 條件式隨機域 | en_US |
| DC.title | 英文郵政地址與鄰近相關資訊擷取之研究 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Application and Extraction of Postal Addresses and Related Information | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |