- ニュース
- 2026年10月7日(水) に産研講演会「Classifying firms' technological capabilities from website text: A longitudinal study of AI adoption in Europe」が開催されます。
2026年10月7日(水) に産研講演会「Classifying firms’ technological capabilities from website text: A longitudinal study of AI adoption in Europe」が開催されます。
Dates
カレンダーに追加1007
WED 2026- Place
- 対面
- Time
- 12:15～13:30
- Posted
- Fri, 18 Sep 2026
「Classifying firms’ technological capabilities from website text: A longitudinal study of AI adoption in Europe」
| 日時 | 2026年10月7日（水）12:15~13:30 |
|---|---|
| 開催方法 | 対面 ＊11号館 4階 第3会議室 |
| 対象 | 学生・教職員・一般 |
| 講演者 | Alexander Kopka 氏（Senior Researcher, Thünen Institute of Rural Economics） |
| 要旨 | AI adoption is expected to drastically impact economic activity in the coming years. However, assessing and tracking firm-level AI adoption in real time remains difficult. Firm-level measures of technological capabilities typically rely on patents, R&D reporting, or surveys, which capture only a narrow and lagged slice of firm activity and tend to skew toward larger firms. We explore whether firm websites, scraped longitudinally and processed with text-mining methods, can serve as a complementary signal of technological capabilities, using AI as a test case. The study draws on the CommonCrawl database, which has captured snapshots of the internet at regular intervals since 2014, combined with firm-level information from ORBIS, to construct a longitudinal EU-wide (EU27+UK) firm-level website text database for classifying (stated) technological capabilities. To classify, we plan a two-stage approach: relevant paragraphs are first identified through a dictionary-based keyword search and then classified using a supervised Sentence-BERT model.
We address two research questions: (a) whether firm website texts are a useful indicator of technological capabilities and how they compare with well-established measures, and (b) how AI adoption is descriptively distributed across the EU. While research on website data for classifying technological capabilities is still young, it tends to (a) focus on individual countries and (b) neglect the potential weaknesses of the indicator itself. This study therefore aims to deliver both a methodological contribution, assessing what a website-based indicator actually captures, and a data contribution, in the form of a longitudinal database of AI capabilities for European firms. |
| 世話人 | 三橋 平（早稲田大学商学学術院 教授） |
| 参加申し込み方法 | 参加はこちらからお申込みください。※10月5日（月）17:00締切 |
| 共催 | 早稲田大学商学部・産業経営研究所・山野井 順一分科会 |
- Tags
- イベント