معرفی شرکت ها
embedded-topic-model-1.0.2
تبلیغات ما
مشتریان به طور فزاینده ای آنلاین هستند. تبلیغات می تواند به آنها کمک کند تا کسب و کار شما را پیدا کنند.
مشاهده بیشتر
تبلیغات ما
مشتریان به طور فزاینده ای آنلاین هستند. تبلیغات می تواند به آنها کمک کند تا کسب و کار شما را پیدا کنند.
مشاهده بیشتر
تبلیغات ما
مشتریان به طور فزاینده ای آنلاین هستند. تبلیغات می تواند به آنها کمک کند تا کسب و کار شما را پیدا کنند.
مشاهده بیشتر
تبلیغات ما
مشتریان به طور فزاینده ای آنلاین هستند. تبلیغات می تواند به آنها کمک کند تا کسب و کار شما را پیدا کنند.
مشاهده بیشتر
تبلیغات ما
مشتریان به طور فزاینده ای آنلاین هستند. تبلیغات می تواند به آنها کمک کند تا کسب و کار شما را پیدا کنند.
مشاهده بیشترتوضیحات
A package to run embedded topic modelling
| ویژگی | مقدار |
|---|---|
| سیستم عامل | - |
| نام فایل | embedded-topic-model-1.0.2 |
| نام | embedded-topic-model |
| نسخه کتابخانه | 1.0.2 |
| نگهدارنده | [] |
| ایمیل نگهدارنده | [] |
| نویسنده | Luiz F. Matos |
| ایمیل نویسنده | lfmatosmelo@id.uff.br |
| آدرس صفحه اصلی | https://github.com/lffloyd/embedded-topic-model |
| آدرس اینترنتی | https://pypi.org/project/embedded-topic-model/ |
| مجوز | MIT license |
# Embedded Topic Model
[](https://badge.fury.io/py/embedded-topic-model)
[](https://github.com/lffloyd/embedded-topic-model/actions)
[](https://github.com/lffloyd/embedded-topic-model/blob/main/LICENSE)
This package was made to easily run embedded topic modelling on a given corpus.
ETM is a topic model that marries the probabilistic topic modelling of Latent Dirichlet Allocation with the
contextual information brought by word embeddings-most specifically, word2vec. ETM models topics as points
in the word embedding space, arranging together topics and words with similar context.
As such, ETM can either learn word embeddings alongside topics, or be given pretrained embeddings to discover
the topic patterns on the corpus.
ETM was originally published by Adji B. Dieng, Francisco J. R. Ruiz, and David M. Blei on a article titled ["Topic Modeling in Embedding Spaces"](https://arxiv.org/abs/1907.04907) in 2019. This code is an adaptation of the [original](https://github.com/adjidieng/ETM) provided with the article. Most of the original code was kept here, with some changes here and there, mostly for ease of usage.
With the tools provided here, you can run ETM on your dataset using simple steps.
# Installation
You can install the package using ```pip``` by running: ```pip install -U embedded_topic_model```
# Usage
To use ETM on your corpus, you must first preprocess the documents into a format understandable by the model.
This package has a quick-use preprocessing script. The only requirement is that the corpus must be composed
by a list of strings, where each string corresponds to a document in the corpus.
You can preprocess your corpus as follows:
```python
from embedded_topic_model.utils import preprocessing
import json
# Loading a dataset in JSON format. As said, documents must be composed by string sentences
corpus_file = 'datasets/example_dataset.json'
documents_raw = json.load(open(dataset, 'r'))
documents = [document['body'] for document in documents_raw]
# Preprocessing the dataset
vocabulary, train_dataset, _, = preprocessing.create_etm_datasets(
documents,
min_df=0.01,
max_df=0.75,
train_size=0.85,
)
```
Then, you can train word2vec embeddings to use with the ETM model. This is optional, and if you're not interested
on training your embeddings, you can either pass a pretrained word2vec embeddings file for ETM or learn the embeddings
using ETM itself. If you want ETM to learn its word embeddings, just pass ```train_embeddings=True``` as an instance parameter.
To pretrain the embeddings, you can do the following:
```python
from embedded_topic_model.utils import embedding
# Training word2vec embeddings
embeddings_mapping = embedding.create_word2vec_embedding_from_dataset(documents)
```
To create and fit the model using the training data, execute:
```python
from embedded_topic_model.models.etm import ETM
# Training an ETM instance
etm_instance = ETM(
vocabulary,
embeddings=embeddings_mapping, # You can pass here the path to a word2vec file or
# a KeyedVectors instance
num_topics=8,
epochs=300,
debug_mode=True,
train_embeddings=False, # Optional. If True, ETM will learn word embeddings jointly with
# topic embeddings. By default, is False. If 'embeddings' argument
# is being passed, this argument must not be True
)
etm_instance.fit(train_dataset)
```
Also, to obtain the topics, topic coherence or topic diversity of the model, you can do as follows:
```python
topics = etm_instance.get_topics(20)
topic_coherence = etm_instance.get_topic_coherence()
topic_diversity = etm_instance.get_topic_diversity()
```
# Citation
To cite ETM, use the original article's citation:
```
@article{dieng2019topic,
title = {Topic modeling in embedding spaces},
author = {Dieng, Adji B and Ruiz, Francisco J R and Blei, David M},
journal = {arXiv preprint arXiv: 1907.04907},
year = {2019}
}
```
# Acknowledgements
Credits given to Adji B. Dieng, Francisco J. R. Ruiz, and David M. Blei for the original work.
# License
Licensed under [MIT](LICENSE) license.
# Changelog
This changelog was inspired by the [keep-a-changelog](https://github.com/olivierlacan/keep-a-changelog) project and follows [semantic versioning](https://semver.org).
## [1.0.2] - 2021-06-23
### Changed
- deactivates debug mode by default
- documents get_most_similar_words method
## [1.0.1] - 2021-02-15
### Changed
- optimizes original word2vec TXT file input for model training
- updates README.md
## [1.0.0] - 2021-02-15
### Added
- adds support for original word2vec pretrained embeddings files on both formats (BIN/TXT)
### Changed
- optimizes handling of gensim's word2vec mapping file for better memory usage
## [0.1.1] - 2021-02-01
### Added
- support for python 3.6
## [0.1.0] - 2021-02-01
### Added
- ETM training with partially tested support for original ETM features.
- ETM corpus preprocessing scripts - including word2vec embeddings training - adapted from the original code.
- adds methods to retrieve document-topic and topic-word probability distributions from the trained model.
- adds docstrings for tested API methods.
- adds unit and integration tests for ETM and preprocessing scripts.
نیازمندی
| مقدار | نام |
|---|---|
| ==3.8.3 | gensim |
| ==3.5 | nltk |
| ==1.19.5 | numpy |
| ==0.23.2 | scikit-learn |
| ==1.5.2 | scipy |
| ==1.6.0 | torch |
زبان مورد نیاز
| مقدار | نام |
|---|---|
| >=3.6 | Python |
نحوه نصب
نصب پکیج whl embedded-topic-model-1.0.2:
pip install embedded-topic-model-1.0.2.whl
نصب پکیج tar.gz embedded-topic-model-1.0.2:
pip install embedded-topic-model-1.0.2.tar.gz