政府開放資料之擷取、分析與融合以建置智慧化API服務

本論文整合文字分析與資料探勘技術，分析政府開放資料，建立5W Knowledge Tree，開發智慧化概念式API服務。隨著政府資料開放平台的建立時間越久，各部會定期上傳各類型的開放資料集累積越多。目前，政府開放資料平台上擁有多達18種服務分類，與數十種不同檔案格式被公開釋出。使得平台擁有潛在價值龐大的開放資料資源，可以供民眾運用與開發服務。然而，在政府資料開放平台上，從分散的資料集尋找所需的答案並不容易，尤其開放資料多數無法直接使用，且資料集之間無法建立關連，更遑論對於不諳程式開發的民眾來說，要使用開放資料更加困難。若能融合不同部門開放資料集，分析資料內涵5W的結構與關聯、挖掘資料集欄位屬性、建立關鍵字與物件之間的關聯，就可能整合與連結不同開放資料集，成為更易於使用者與開發人員使用的五顆星等開放資料。因此，本論文以「科技部」與「教育部」開放資料為案例，整合兩個部門的開放資料，開發出智慧化API服務，提供連一般民眾也能以關鍵字概念查詢資料，並且以表單資料或視覺化圖表統計資料。

關鍵字

文本探勘；資料探勘； Restful API ； LOD ； RDF

並列摘要

This paper integrates the technology of text analysis and data exploration, analyses the open data set provided by the government, establishes 5W Knowledge Tree, and develops intelligent conceptual API service. With the establishment of the platform for government open data, ministries will regularly upload various types of open data sets and accumulate more. Currently, there are up to 18 service categories on the government's open data platform, and dozens of different file formats are publicly released. The platform has a potentially valuable open data resources, can be used and developed for the public. However, on government data open platforms, it is not easy to figure out the answers needed from scattered data sets, especially when most open data are not directly used and there is no connection between data sets, not to mention it is more difficult for people who do not develop programs to use open data. If we can fusion the open datasets that provided from different departments, analyze the structure and relation of the data connotation, text mining for knowledge triple extraction of the datasets, and establish the association between keywords and objects. It is possible to integrate and link different open data sets to become the five-star open data that is easier for users and developers to use. Therefore, this paper takes the "Ministry of Science and Technology" and "Ministry of Education" open data as a case study, integrates the open data of the two departments, develops intelligent API services, provides the general public can query data with Conceptualized API Path, and to form data or visual chart statistics.