Table 1 Comparison of TF-IDF based on original dataset and expanded dataset (In [%]).

From: LLM-based intelligent Q&A system for railway locomotive maintenance standardization

Dataset type/Feature word

Replace

Normal

Test

Inspection

Driver

Malfunction

Storage

Operation

Seat

Good

Original dataset

54.11

50.68

37.74

33.82

15.85

15.23

14.19

10.83

8.47

7.70

Expanded dataset

51.29

62.03

29.78

26.88

14.16

13.26

11.21

9.83

6.68

6.10