respapers_topics

respapers_topics

This is a BERTopic model. BERTopic is a flexible and modular topic modeling framework that allows for the generation of easily interpretable topics from large datasets.

This pre-trained model was built to demonstrate the use of representation model inspired on KeyBERT to be use within BERTopic.

This model was trained on ~30000 Research Papers abstracts with the KeyBERTInspired representation method (bertopic.representation). The dataset was downloaded from kaggle, with the two subsets (test and train) being merged into a single dataset.

To access the complete code, you can vist this tutorial on my GitHub page: ResPapers

Usage

To use this model, please install BERTopic:

pip install -U bertopic

You can use the model as follows:

from bertopic import BERTopic
topic_model = BERTopic.load("CCatalao/respapers_topics")

topic_model.get_topic_info()

To view the KeyBERT inspired topic representation please use the following:

>>> topic_model.get_topic(0, full=True)
{'Main': [['spin', 0.01852648864225281],
  ['magnetic', 0.015019436257929909],
  ['phase', 0.013081733986038124],
  ['quantum', 0.012942253723133639],
  ['temperature', 0.012591407440537158],
  ['states', 0.011025582290837643],
  ['field', 0.010954775154251296],
  ['electron', 0.010168708734803916],
  ['transition', 0.009728560280580357],
  ['energy', 0.00937042795113575]],
 'KeyBERTInspired': [['quantum', 0.4072583317756653],
  ['phase transition', 0.35542067885398865],
  ['lattice', 0.34462833404541016],
  ['spin', 0.3268473744392395],
  ['magnetic', 0.3024371564388275],
  ['magnetization', 0.2868726849555969],
  ['phases', 0.27178525924682617],
  ['fermi', 0.26290175318717957],
  ['electron', 0.25709500908851624],
  ['phase', 0.23375216126441956]]}

Topic overview

• Number of topics: 112
• Number of training documents: 29961

<details> <summary>Click here for an overview of all topics.</summary> </details>

Training Procedure

The model was trained as follows:

from bertopic import BERTopic
from sklearn.feature_extraction.text import CountVectorizer
from bertopic.representation import KeyBERTInspired

from sentence_transformers import SentenceTransformer
from umap import UMAP
from hdbscan import HDBSCAN

# Prepre sub-models
embedding_model = SentenceTransformer('sentence-transformers/all-mpnet-base-v2')
umap_model = UMAP(n_components=5, n_neighbors=50, random_state=42, metric="cosine", verbose=True)
hdbscan_model = HDBSCAN(min_samples=20, gen_min_span_tree=True, prediction_data=False, min_cluster_size=20)
vectorizer_model = CountVectorizer(stop_words="english", ngram_range=(1, 3), min_df=5)

# Representation models
representation_models = {"KeyBERTInspired": KeyBERTInspired()}

# Fit BERTopic
topic_model = BERTopic(
                umap_model=umap_model,
                hdbscan_model=hdbscan_model,
                vectorizer_model=vectorizer_model,
                representation_model=representation_models,
                min_topic_size= 10,
                n_gram_range= (1, 1),
                nr_topics=None,
                seed_topic_list=None,
                top_n_words=10,
                calculate_probabilities=False,
                language=None,
                verbose = True
).fit(docs)

Training hyperparameters

• calculate_probabilities: False
• language: None
• low_memory: False
• min_topic_size: 10
• n_gram_range: (1, 1)
• nr_topics: None
• seed_topic_list: None
• top_n_words: 10
• verbose: True

Framework versions

• Numpy: 1.22.4
• HDBSCAN: 0.8.33
• UMAP: 0.5.3
• Pandas: 1.5.3
• Scikit-Learn: 1.2.2
• Sentence-transformers: 2.2.2
• Transformers: 4.29.2
• Numba: 0.56.4
• Plotly: 5.13.1
• Python: 3.10.11