Accesso libero

The Application of Artificial Intelligence Big Model Technology in Mental Health Education Work in Colleges and Universities

Ying Li

Ying Li

Mental Health Education Center, Henan Technical InstituteZhengzhou, China
Cerca questo autore su
Sciendo|Google Scholar
Li, Ying
 e 
Chenjian Xu

Chenjian Xu

Informatization Center, ShanDong Business InstituteYantai, China
Cerca questo autore su
Sciendo|Google Scholar
Xu, Chenjian
  
24 set 2025
The Application of Artificial Intelligence Big Model Technology in Mental Health Education Work in Colleges and Universities's Cover Image
INFORMAZIONI SU QUESTO ARTICOLO
Articolo precedente
Articolo Successivo
Cita
Scarica la copertina
Pubblicato online: 24 set 2025
Ricevuto: 14 gen 2025
Accettato: 01 mag 2025
DOI: https://doi.org/10.2478/amns-2025-1019
Parole chiave
© 2025 Ying Li et al., published by Sciendo.
This work is licensed under the Creative Commons Attribution 4.0 International License.

Figure 1.

Rasa’s Architecture Diagram
Rasa’s Architecture Diagram

Figure 2.

Unidirectional paraphrase learning and bi-directional paraphrase learning
Unidirectional paraphrase learning and bi-directional paraphrase learning

Figure 3.

UniLM-Copy mechanism model structure
UniLM-Copy mechanism model structure

Figure 4.

Retrieval performance comparison
Retrieval performance comparison

Figure 5.

Experimental test comparison
Experimental test comparison

Figure 6.

Guidance mode comparison
Guidance mode comparison

Different concurrent response schedules

Concurrency 10 30 50 100 300
Average response time (s) 0.524 0.619 0.837 1.059 1.345
Maximum response time (s) 0.832 0.826 1.139 1.256 1.579
Minimum response time (s) 0.416 0.467 0.630 0.838 1.036

Statistics of ESC Conv data set

Parameter Test set Search set (training set + verification set)
Session count 200 1110
Number of statements 6037 32327
Average number of session rounds 30.95 29.26
Average statement length 15.70 16.55

Automatic evaluation results(units:%)

Model Method BLEU-2 BLEU-4 METEOR Distinct-1 Distinct-2
Mistral +Vanilla 2.37 0.83 12.03 5.24 25.35
+CoT 2.47 0.81 12.38 5.15 25.84
+UniLM-Copy 2.62 1.11 12.42 7.08 36.27
Gemma +Vanilla 1.97 0.76 9.23 6.74 29.17
+CoT 1.93 0.74 9.18 6.53 28.43
+UniLM-Copy 2.11 0.80 10.23 7.17 33.72
Llama2 +Vanilla 2.19 0.75 12.88 3.57 17.38
+CoT 2.09 0.71 12.68 3.62 17.59
+UniLM-Copy 2.22 0.74 13.84 4.23 21.07
Llama3 +Vanilla 2.44 0.94 11.23 5.77 26.04
+CoT 2.52 1.05 10.61 5.77 24.85
+UniLM-Copy 2.68 1.02 11.71 9.26 40.66

Human evaluation results(units:%)

Comparison Evaluation content Win Lose Tie
Baseline method Fluency 43.1 22.1 35.1
Identifiability 65.1 11.8 23.4
Comfort 59.4 12.4 28.5
Suggestion 52.4 16.8 31.1
Totality 60.8 15.1 24.4
Textual method Fluency 49.4 14.4 36.5
Identifiability 63.8 9.4 27.1
Comfort 55.8 21.1 23.4
Suggestion 50.1 15.8 34.4
Totality 58.4 19.8 22.1
Lingua:
Inglese
Frequenza di pubblicazione:
1 volte all'anno
Argomenti della rivista:
Scienze biologiche, Scienze della vita, altro, Matematica, Matematica applicata, Matematica generale, Fisica, Fisica, altro
Feed RSS della rivista