MarieAngeA13
commited on
Commit
•
f71d85d
1
Parent(s):
c8fe516
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,47 @@
|
|
1 |
---
|
2 |
license: bsd-3-clause
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: bsd-3-clause
|
3 |
+
|
4 |
+
language:
|
5 |
+
- en
|
6 |
+
|
7 |
+
tags:
|
8 |
+
- sentiment
|
9 |
+
- bert
|
10 |
+
- sentiment-analysis
|
11 |
+
- transformers
|
12 |
+
|
13 |
+
pipeline_tag: text-classification
|
14 |
---
|
15 |
+
> Authors : GRP209
|
16 |
+
|
17 |
+
# User Comment Sentiment Analysis
|
18 |
+
|
19 |
+
This model aims to analyze user comments on products and extracting the expressed sentiments.
|
20 |
+
|
21 |
+
User ratings on the internet do not always provide detailed qualitative information about their experience.
|
22 |
+
|
23 |
+
Therefore, it is important to go beyond these ratings and extract more insightful information that can help a brand improve their product or service.
|
24 |
+
|
25 |
+
# Objective
|
26 |
+
|
27 |
+
The model utilizes the BERT architecture and is trained on a dataset of user comments with sentiment labels.
|
28 |
+
|
29 |
+
The model is capable of analyzing comments and extracting sentiments such as **positive**, **negative**, or **neutral**.
|
30 |
+
|
31 |
+
# Features
|
32 |
+
|
33 |
+
**Sentiment Classification**: The model can classify user comments into positive, negative, or neutral sentiments, providing an overall indication of the expressed opinion.
|
34 |
+
|
35 |
+
**Improvement Suggestions**: In cases where a comment expresses a negative or neutral sentiment, the model suggests an improved version of the text with a more positive sentiment.
|
36 |
+
This can help businesses understand consumer reactions and identify areas for product or service improvement.
|
37 |
+
|
38 |
+
# Usage
|
39 |
+
|
40 |
+
To use this sentiment analysis system, follow these steps:
|
41 |
+
|
42 |
+
- Install the required dependencies by running the command pip install -r requirements.txt.
|
43 |
+
- Once the training is complete, the best-trained model will be saved in the best_model_state.bin file.
|
44 |
+
- To make predictions on new comments, use the analyze_sentiment(comment_text) function, replacing comment_text with the actual comment text to analyze.
|
45 |
+
- The model will return the sentiment expressed in the comment.
|
46 |
+
- To suggest an improved version of a comment, use the suggest_improved_text(comment_text) function.
|
47 |
+
- If the comment expresses a negative or neutral sentiment, the function will generate an improved version of the text with a more positive sentiment. Otherwise, the original text will be returned without modification.
|