OntoCraft
ICT trends

ICT trends, read as a knowledge graph

When nearly every publication carries AI, counting them tells you nothing. So we look at what arrives together with what. The material is published by Korea's IITP, and we state the sample size and what we could not classify.

IITP · 2021.11 ~ 2026.11 · 553 items

The overview tab shows aggregates only. Pick a year to see the publications. Putting all 553 on one page makes it far too heavy.

Source

Published by 정보통신기획평가원 (IITP)

Obtained via
공공데이터포털 15036878 (등록기관 한국연구재단)
Terms of use
이용허락범위 제한 없음 (공공누리 유형 미표기)
Updated
2026.11.19

553

Publications collected

252

With an outline

553

Source links

231

Without a topic

An outline is the issue list printed at the front of a publication, not the full text. We do not host the documents; we link to the publisher instead.

“Source links” adds the 205 items that carry a PDF URL to the 348 we open on the publisher's detail page.

① Topic relations

What comes with what

A publication carries several topics at once. The lines are how often two topics appeared together; circle size is how many items carry that topic.

AI–반도체, 86 itemsAI–데이터, 77 itemsAI–보안, 46 itemsAI–플랫폼, 41 items데이터–반도체, 36 itemsAI–양자, 29 items데이터–플랫폼, 26 itemsAI–로봇, 25 items보안–반도체, 22 itemsAI–인재, 22 items플랫폼–반도체, 21 items데이터–보안, 21 itemsAI, 236 items236AI반도체, 128 items128반도체데이터, 96 items96데이터플랫폼, 58 items58플랫폼보안, 52 items52보안양자, 43 items43양자로봇, 32 items32로봇인재, 26 items26인재인프라·데이터센터, 13 items13인프라·데이터센터공급망·수출통제, 3 items3공급망·수출통제
Links that appeared only recentlyA thicker line means the two appeared together more often

What changed recently

Grown in share

  • AI · 인프라·데이터센터

    3 items 3.3% → 10 items 10.1%

  • 데이터 · 인프라·데이터센터

    2 items 2.2% → 7 items 7.1%

  • AI · 인재

    4 items 4.3% → 9 items 9.1%

  • AI · 양자

    7 items 7.6% → 12 items 12.1%

  • AI · 데이터

    24 items 26.1% → 30 items 30.3%

ⓘ Only quarters classified above 50% are compared. The earlier window is 24Q2~25Q3 (92 classified); the recent one is 25Q4~26Q4 (99 classified).

ⓘ We compare shares of the items classified in each window, not raw counts. Raw counts cannot tell a new relation apart from more classification.

ⓘ Only the 12 most frequent pairs are drawn; the cut-off is 21 items. Drawing them all turns the picture into a hairball.

ⓘ Each list shows at most 5 entries, largest change first.

ⓘ We do not list what fell. The differences were one or two items, which is sampling noise rather than a trend.

① Topic flow

How many items each topic carried

One bar is one quarter, the rightmost being the latest. The number on the right is the total for the whole window.

The bars count only what carries a topic. Earlier quarters have been gone through less, so a low bar there does not mean the topic was absent. Each quarter shows how much of it is classified, and quarters below 50% are hatched.

  • AI

    236
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 반도체

    128
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 데이터

    96
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 플랫폼

    58
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 보안

    52
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 양자

    43
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 로봇

    32
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 인재

    26
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 인프라·데이터센터

    13
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4
  • 공급망·수출통제

    3
    21Q422Q122Q222Q322Q423Q123Q223Q323Q424Q124Q224Q324Q425Q125Q225Q325Q426Q126Q226Q326Q4

ⓘ Each quarter holds 26.3 items on average. The sample is small, so read the movement between quarters as a direction, not a measurement.

②

Rising topics

  • 반도체26Q3 4 → 26Q4 6
  • 데이터26Q3 2 → 26Q4 4
  • 플랫폼26Q3 0 → 26Q4 2

ⓘ This shows how often the publisher wrote about a topic. It does not say the market grew. The latest quarter is also still in progress, so the comparison is early.

③

Terms seen for the first time

  • 이기종26Q4
  • 공개해26Q4
  • 전략산업26Q4
  • 다층26Q4
  • 스케일링26Q4
  • 체코26Q4
  • 통보26Q4
  • 시정26Q4
  • 무엇을26Q4
  • 제공해26Q4
  • APACHE26Q4
  • 병행해26Q4
  • 총운영비26Q4
  • 신질생산력26Q4
  • 토큰당26Q4
  • 개발동향26Q4
  • L226Q4
  • L326Q4

ⓘ Terms that first appeared in the quarter shown. They are not yet settled topics.

④ No topic attached

231 items
  • Source gave us nothing190The title is only the series name and issue number, and no outline arrived.
  • Outside the dictionary41There is something to go on, but our topic dictionary does not know the term.
Show the 41 outside the dictionary
21Q4 922Q1 1422Q2 1622Q3 1522Q4 1523Q1 2023Q2 2523Q3 2623Q4 2424Q1 924Q2 1224Q3 1124Q4 2625Q1 825Q2 025Q3 025Q4 126Q1 026Q2 026Q3 026Q4 0

Placed as relations, it reads differently

The same material tells a different story once you place what connects to what. Weaving scattered documents and data into something you can ask questions of is the work we do. What we wrote ourselves about ontologies and knowledge graphs is in Insights.