Open Access

Improving Mandarin Emotional Vocabulary in Application Writing Using Deep Learning Models

  
Sep 26, 2025

Cite
Download Cover

Figure 1.

Emotion dictionary construction process diagram
Emotion dictionary construction process diagram

Figure 2.

Classification of emotion words
Classification of emotion words

Figure 3.

ECBOW model diagram
ECBOW model diagram

Figure 4.

Binary tree of semantic dependency structure
Binary tree of semantic dependency structure

Figure 5.

Extraction of binary semantic dependency path feature map
Extraction of binary semantic dependency path feature map

Figure 6.

Constructing emotion word classifier using bidirectional LSTM neural network
Constructing emotion word classifier using bidirectional LSTM neural network

Figure 7.

The characteristics of writing in five schools are different
The characteristics of writing in five schools are different

Figure 8.

Positive, negative and negative to emotional decomposition
Positive, negative and negative to emotional decomposition

Figure 9.

The top 20 positive emotional entries in the writing text of five schools
The top 20 positive emotional entries in the writing text of five schools

Figure 10.

The top 20 negative emotional entries in the writing text of five schools
The top 20 negative emotional entries in the writing text of five schools

Figure 11.

The application text is based on the comparison
The application text is based on the comparison

The emotional word calculates the sample

Candidate word Forward frequency Negative frequency Threshold value=1 Threshold value=2 Threshold value=3 Emotional intensity
Very General 34 43 -1.2438 -2.1507 -3.0974 -0.6727
Cure 846 18 2.8314 2.0781 1.1103 0.9815
Good 693 14 2.9508 1.5632 0.6642 0.9286
Generally 34 12 0.4187 -0.7638 -1.5323 -0.1298
Chicken Soup 30 61 -1.6404 -2.466 -3.535 -0.6016
Embarrassing 213 46 0.3502 -0.5849 -1.4594 0.6812
Dry Goods 33 27 -0.8731 -2.0351 -3.0566 -0.4729
Brain Hole 688 9 2.9753 2.2108 1.2001 1.1169

A conditional filtering process recording table based on rules

Elimination rule Remaining candidate affective words Positive word Negative term
Initial 929229 -- --
(1) 90307 69943 20330
(2) 77557 66979 10585
(3) 65684 61547 8153
(4)~(6) 62930 55498 7405
(7)~(9) 59383 52536 6804

The emotional dictionary is the same as the result

Categories Dictionary size Recall rate Accuracy rate F1
English language dictionary 59115 91.34% 78.48% 83.73%
Dalian polytechnic emotional dictionary 27463 37.54% 70.38% 48.09%
The hownet emotional dictionary 8733 35.34% 75.76% 46.92%

The statistical phase of the statistical phase of the word frequency

Typo 1 star 2 star 4 star 5 star Forward Negative direction Total quantity
Quantity/piece 61986 121715 973369 1226827 2285165 181825 4850887

Vocabulary score statistics

Vocabulary Book number Number of users 1 star 2 star 3 star 4 star 5 star
Introduction 1207 1234 4 29 363 598 389
Good 1984 2217 21 57 370 1150 841
Unintelligible 1119 1315 250 349 359 281 165
…… 637 577 25 53 349 155 86
Look at it 814 1039 24 115 351 354 328
Incomprehension 1207 1234 -1 29 363 598 389

High frequency emotional term top30

Positive emotion Good, good, good, great, great, great love, great love, great love, good, still, move, a little meaning, a little meaning, the recommendation, great, the shock, is worth reading, is worth reading, touching, is worth seeing, good, warm, very interesting
Negative emotion Generally, I don’t understand, garbage, don’t like, don’t look, generally, don’t understand, see not to go, copy, still, I don’t know, I don’t read it, I don’t read it, I don’t know, it’s very general, no sense, bad, general, melodramatic, I don’t understand, I can’t read it. No meaning, tail, disappointment, Have seen, What the devil, Forget, forget, read and don’t understand, speechless

Writing text feature statistics

Institutional type Text number Basic statistics Grammatical index
Total word The words are all words It’s greater than the ratio of 6 letters Proportional pronoun (%) Proportional pronoun (%) Proportional pronoun (%) Proportional pronoun (%)
Key university 86574 220.08 21.44 27.57 10.94 7.53 14.16 5.83
School of our school 602045 193.14 25.61 25.41 12.38 9.63 13.95 7.19
Vocational school 40405 188.85 26.43 30.35 7.41 7.25 13.74 9.16
High school 4042 140.66 23.66 22.58 9.67 6.14 14.62 3.99
Junior high school 1463 130.33 21.14 14.54 11.76 9.38 12.04 6.38

Experimental effect evaluation confusion matrix

Forecast
Correctness Errors
Actual Correctness TP FN
Errors FP TN
Language:
English