Texts, provenance, and access
The course uses prepared, documented corpora so that class time can focus on analytical decisions rather than large-scale data collection and cleaning.
A corpus is a collection of documents that you can use for analysis. In this course, a document might be a presidential speech or a social-media post. Each document has text and may have metadata, such as its author, date, and type.
What belongs in the collection affects the questions it can help answer. For each course corpus, check who produced the documents and why, which documents were included, what each row represents, and what is missing. The corpus card also explains where the material came from and how you may use it.
Primary focus
Most demonstrations will use prepared selections of South Korean presidential speeches. Speeches provide substantial documents and useful metadata for comparing speakers, periods, speech types, and political contexts.
A difference between presidents may reflect the speech type or period. The audience, document length, and selection rules may also account for the difference.
Limited use
A smaller number of exercises may use prepared selections of presidential social-media posts, principally from Moon Jae-in’s account. Tweets are shorter, platform-specific, and shaped by posting and collection practices.
Engagement measures do not measure public opinion. An account’s posts also cover only part of a president’s public communication.
Session 5
Sentences from the minutes of the National Assembly, 2000 to 2026, that name North Korea or Japan. Each sentence carries its speaker, party, committee, and date, so you can compare how ruling and opposition members speak about each country.
A sentence is cut from a longer speech, and members speak for an audience. The minutes record what was said, not what members think.
The speeches are used in Sessions 2–4 and 6. Session 5 uses the Assembly sentences, and the tweets in its optional preparation. Other materials may be introduced for a specific lesson. Classroom use does not grant permission for independent reuse.
Sessions 2–6 · Orange
Orange does not preprocess Korean text. The script splits each eojeol into morphemes, tags parts of speech, and keeps nouns. Load it in a Python Script widget with More › Import Script from File, then click Run. The first run installs Kiwi and needs internet. Load the stopword file in Preprocess Text under Filtering › Stopwords.
The final hackathon uses a prepared menu of course corpora. Each option will include the following materials.
The menu is released with the hackathon brief on Oct. 30. Groups choose from the approved menu. They do not collect or clean a new dataset during the timed event. The assessment covers the analysis and its verification. It also covers how well the group communicates the result.
Keep course repositories private unless instructed otherwise. Never commit personal data, credentials, restricted files, or a full corpus simply because you can access it. If a corpus card says “metadata only” or “local use only,” those terms apply to every member of the group.