File size: 10,003 Bytes
af2ba21 123f56f af2ba21 123f56f 5a4beae 3c4d5ef 123f56f 5804ce1 123f56f 5804ce1 123f56f 3c4d5ef 123f56f 3c4d5ef 5a4beae 123f56f 01d8951 123f56f af2ba21 123f56f 01d8951 123f56f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 | ---
library_name: transformers
license: mit
language:
- en
base_model:
- google-bert/bert-base-uncased
pipeline_tag: text-classification
tags:
- PyTorch
- multi-class-classification
---
This is Bert-base-uncased model fine-tuned for topic classification of therapist remarks in psychotherapeutic contexts. The task is a multi-class classification with the following labels:
```python
id2label = {
0: 'Relationship with a Female Figure',
1: 'Experiencing and Articulating Emotions',
2: 'Work and Personal Fulfillment',
3: 'Relationship with a Male Figure',
4: 'Meaning-Making and Conceptualization',
5: 'Education and Personal Development',
6: 'Experience of Fear and Anxiety',
7: 'Sexual Relationships and Identity',
8: 'Expression and Communication Difficulties',
9: 'Self-Acceptance and Social Relationships',
10: 'Time Perception and Management',
11: 'Cyclical Patterns of Experience',
12: 'Therapy Experience and Reflection',
13: 'Positive Emotional Experience',
14: 'Mother-Child Relationship Dynamics',
15: 'Creativity and Emotional Expression',
16: 'Humor and Cognitive Reframing',
17: 'Emotional Disturbance and Regulation',
18: 'Expressions of Certainty',
19: 'Guilt and Responsibility',
20: 'Self-Reflection on Maturity',
21: 'Sleep-Wake Cycle and Daily Rhythm',
22: 'Life Orientation and Authenticity',
23: 'Home and Attachment-Autonomy Conflict',
24: 'Parenting and Family Responsibility',
25: 'Internal Conflict and Emotional Regulation',
26: 'Friendship Dynamics',
27: 'Reflection and Awareness of Unconsidered Thought',
28: 'Reading as Cognitive and Reflective Engagement',
29: 'Processes of Personal Change',
30: 'Emotional Pain and Hurt',
31: 'Belief, Faith, and Existential Doubt',
32: 'Evaluation of Neurotic and Healthy States',
33: 'Emotional Distress and Negative Evaluation',
34: 'Experience and Regulation of Anger',
35: 'Positive Emotional Evaluation',
36: 'Dream Experience and Symbolic Processing',
37: 'Father-Child Relationship Dynamics',
38: 'Problem Framing and Self-Resolution',
39: 'Personal Learning Experiences',
40: 'Racial Identity and Intergroup Relations',
41: 'Healthcare Interactions and Medical Authority',
42: 'Uncertainty about Correctness',
43: 'Evaluation of Marriage',
44: 'Marital Relationship and Communication',
45: 'Self-Worth and Capacity for Love',
46: 'Self-Expression Through Appearance'}
```
## Usage
```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("AIPsy/bert-base-client-topic-classification-eng")
model = AutoModelForSequenceClassification.from_pretrained("AIPsy/bert-base-client-topic-classification-eng")
text = "That in essense it's right; it's responsibility."
encoding = tokenizer(
text,
truncation=True,
padding="max_length",
return_tensors="pt"
)
output = model(encoding['input_ids'], encoding['attention_mask']).logits
result = np.argmax(output.detach().numpy(), axis=-1)
print(id2label[result[0]])
'Guilt and Responsibility'
```
## Dataset
The source material was the recordings of psychotherapeutic sessions posted on YouTube in the public domain. After conducting speaker diarization and transcription of the recordings 15324 items (sentences) were obtained.
## Recommendations
Given the broad context of issues discussed in psychotherapeutic sessions, the authors believe that this model can be used to analyze human communication in general.
## Metrics
Score metrics of trained model
|Sample|F1 macro|
|:-|:-:|
|Test|0.76|
|Validation|0.75|
|Train|0.94|
F1 score metrics for test sample across categories
|ID|Topic|Precision|Recall|F1|
|:-:|:-|:-:|:-:|:-:|
| 0 | Relationship with a Female Figure | 0.91 | 0.97 | 0.94 | 232 |
| 1 | Experiencing and Articulating Emotions | 0.77 | 0.82 | 0.79 | 238 |
| 2 | Work and Personal Fulfillment | 0.75 | 0.77 | 0.76 | 235 |
| 3 | Relationship with a Male Figure | 0.86 | 0.91 | 0.88 | 150 |
| 4 | Meaning-Making and Conceptualization | 0.79 | 0.81 | 0.80 | 277 |
| 5 | Education and Personal Development | 0.80 | 0.73 | 0.77 | 83 |
| 6 | Experience of Fear and Anxiety | 0.97 | 0.75 | 0.85 | 80 |
| 7 | Sexual Relationships and Identity | 0.76 | 0.80 | 0.78 | 66 |
| 8 | Expression and Communication Difficulties | 0.70 | 0.83 | 0.76 | 143 |
| 9 | Self-Acceptance and Social Relationships | 0.58 | 0.81 | 0.68 | 221 |
| 10 | Time Perception and Management | 0.85 | 0.71 | 0.77 | 96 |
| 11 | Cyclical Patterns of Experience | 0.79 | 0.65 | 0.71 | 175 |
| 12 | Therapy Experience and Reflection | 0.86 | 0.64 | 0.73 | 58 |
| 13 | Positive Emotional Experience | 0.80 | 0.63 | 0.71 | 78 |
| 14 | Mother-Child Relationship Dynamics | 0.82 | 0.75 | 0.79 | 61 |
| 15 | Creativity and Emotional Expression | 0.64 | 0.73 | 0.68 | 37 |
| 16 | Humor and Cognitive Reframing | 0.60 | 0.63 | 0.62 | 57 |
| 17 | Emotional Disturbance and Regulation | 0.74 | 0.71 | 0.72 | 193 |
| 18 | Expressions of Certainty | 0.96 | 0.91 | 0.94 | 207 |
| 19 | Guilt and Responsibility | 0.62 | 0.59 | 0.60 | 75 |
| 20 | Self-Reflection on Maturity | 0.71 | 0.80 | 0.75 | 45 |
| 21 | Sleep-Wake Cycle and Daily Rhythm | 0.78 | 0.61 | 0.68 | 41 |
| 22 | Life Orientation and Authenticity | 0.75 | 0.70 | 0.72 | 82 |
| 23 | Home and Attachment-Autonomy Conflict | 0.69 | 0.72 | 0.70 | 46 |
| 24 | Parenting and Family Responsibility | 0.79 | 0.70 | 0.74 | 53 |
| 25 | Internal Conflict and Emotional Regulation | 0.87 | 0.68 | 0.76 | 38 |
| 26 | Friendship Dynamics | 0.56 | 0.80 | 0.66 | 59 |
| 27 | Reflection and Awareness of Unconsidered Thought | 0.80 | 0.70 | 0.75 | 94 |
| 28 | Reading as Cognitive and Reflective Engagement | 0.62 | 0.86 | 0.72 | 35 |
| 29 | Processes of Personal Change | 0.69 | 0.69 | 0.69 | 39 |
| 30 | Emotional Pain and Hurt | 0.83 | 0.59 | 0.69 | 32 |
| 31 | Belief, Faith, and Existential Doubt | 0.76 | 0.91 | 0.83 | 32 |
| 32 | Evaluation of Neurotic and Healthy States | 0.75 | 0.75 | 0.75 | 32 |
| 33 | Emotional Distress and Negative Evaluation | 0.65 | 0.58 | 0.61 | 53 |
| 34 | Experience and Regulation of Anger | 0.61 | 0.70 | 0.65 | 33 |
| 35 | Positive Emotional Evaluation | 0.83 | 0.69 | 0.76 | 49 |
| 36 | Dream Experience and Symbolic Processing | 0.94 | 0.84 | 0.89 | 19 |
| 37 | Father-Child Relationship Dynamics | 0.81 | 0.83 | 0.82 | 30 |
| 38 | Problem Framing and Self-Resolution | 0.78 | 0.62 | 0.69 | 40 |
| 39 | Personal Learning Experiences | 0.59 | 0.58 | 0.58 | 33 |
| 40 | Racial Identity and Intergroup Relations | 1.00 | 0.73 | 0.85 | 15 |
| 41 | Healthcare Interactions and Medical Authority | 0.84 | 0.84 | 0.84 | 25 |
| 42 | Uncertainty about Correctness | 0.81 | 0.65 | 0.72 | 120 |
| 43 | Evaluation of Marriage | 0.79 | 0.86 | 0.83 | 22 |
| 44 | Marital Relationship and Communication | 0.76 | 0.62 | 0.68 | 21 |
| 45 | Self-Worth and Capacity for Love | 0.93 | 0.64 | 0.76 | 22 |
| 46 | Self-Expression Through Appearance | 0.62 | 0.64 | 0.63 | 25 |
## Citation
- **Papers:** Vanin A, Bolshev V and Panfilova A (2025) Psychotherapist remarks’ ML classifier: insights from LLM and topic modeling application. Front. Psychiatry 16:1608163. doi: 10.3389/fpsyt.2025.1608163. <https://doi.org/10.3389/fpsyt.2025.1608163>
- **Developed by:** @myentity, @VadZhen, @Alek123
- **License:** MIT
**BibTeX:**
```
@ARTICLE{10.3389/fpsyt.2025.1608163,
AUTHOR={Vanin, Alexander and Bolshev, Vadim and Panfilova, Anastasia },
TITLE={Psychotherapist remarks’ ML classifier: insights from LLM and topic modeling application},
JOURNAL={Frontiers in Psychiatry},
VOLUME={Volume 16 - 2025},
YEAR={2025},
URL={https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2025.1608163},
DOI={10.3389/fpsyt.2025.1608163},
ISSN={1664-0640},
``` |