Trends
Emerging terms in paper titles across the indexed venues
2026 is still being ingested. At least one venue's 2026 paper count is far below its 2025 count - whether because it hasn't published yet or its proceedings are only partly scraped - so counts below are lower across the board, especially for terms concentrated in that venue. Read "Declining" as "not indexed yet", not as a shrinking research area.
Rising (2026)
large reasoning
11 → 35 titles
+218%
dimensional
15 → 39 titles
+160%
feed-forward
16 → 36 titles
+125%
diffusion language
12 → 24 titles
+100%
verifiable
16 → 31 titles
+94%
geometry-aware
15 → 29 titles
+93%
spatial reasoning
17 → 32 titles
+88%
gui agents
13 → 24 titles
+85%
radar
17 → 30 titles
+76%
proposal
12 → 21 titles
+75%
visual token
12 → 21 titles
+75%
uav
14 → 24 titles
+71%
humor
20 → 34 titles
+70%
video reasoning
10 → 17 titles
+70%
agentic
82 → 139 titles
+70%
abductive
13 → 22 titles
+69%
underwater
16 → 27 titles
+69%
semantic alignment
13 → 21 titles
+62%
geo-localization
10 → 16 titles
+60%
clarity
12 → 19 titles
+58%
No 2025 baseline (2026)
Terms with no 2025 titles, so there is no growth rate to compute. Ranked by 2026 volume. Most are returning after a gap year rather than appearing for the first time.
dimensional aspect-based sentiment
24 titles
narrative similarity
23 titles
humor generation
21 titles
marker extraction
16 titles
story similarity
12 titles
extraction and detection
11 titles
parseme
11 titles
narrative story
10 titles
emotional valence
7 titles
everyday knowledge
7 titles
task dimensional
7 titles
adaptive memory
6 titles
evasion detection
6 titles
hyperscale
6 titles
mwe
6 titles
3d foundation
5 titles
against ai
5 titles
codebert
5 titles
dimensional sentiment
5 titles
document layout
5 titles
Declining (2026)
shared task
218 → 22 titles
-90%
state space models
49 → 5 titles
-90%
bangla
55 → 6 titles
-89%
simplification
44 → 5 titles
-89%
acoustic
69 → 8 titles
-88%
grammatical error
40 → 5 titles
-88%
speech recognition
128 → 16 titles
-88%
position
127 → 16 titles
-87%
automatic speech
39 → 5 titles
-87%
language modeling
62 → 8 titles
-87%