With the advancements in deep learning and Natural Language Processing (NLP), the methods for processing and analyzing text data have diversified. In this post, I will explain in detail how to retrieve COVID-19 information related to Moderna from Wikipedia using the Hugging Face library. Hugging Face Transformers provide many pretrained models widely used in NLP tasks, allowing users to easily analyze text data.<\/p>\n
Hugging Face is a platform that provides various pretrained models and tools to facilitate easy use of NLP models. In particular, the Transformers<\/strong> library includes various state-of-the-art transformer models, such as BERT, GPT-2, and T5, enabling users to perform natural language processing tasks more easily.<\/p>\n To use the Hugging Face library, you need to set up a Python environment. You can install the required libraries with the command below.<\/p>\n To retrieve information related to Moderna and COVID-19 from Wikipedia, we will use wikipedia-api<\/strong>. This library provides the ability to easily search Wikipedia pages and fetch their content.<\/p>\n The code below is a simple example that searches for information about Moderna and prints the content.<\/p>\n By running the above code, you can retrieve content from the Wikipedia page of Moderna. Now, let’s check for additional information related to COVID-19.<\/p>\n Similarly, the code to retrieve information about COVID-19 from Wikipedia is as follows.<\/p>\n The text retrieved from Wikipedia must go through a preprocessing step before being inputted into the model. This process involves removing unnecessary characters or symbols and organizing the necessary information.<\/p>\n The code below shows how to remove unnecessary characters from the retrieved text and organize it in a list format.<\/p>\n To analyze the retrieved data, we can use Hugging Face Transformers. Here, we will look at how to input the preprocessed text into the BERT model to extract features.<\/p>\n Let’s use the Hugging Face BERT model to extract features from the preprocessed text. Please refer to the code below.<\/p>\n Now we will create a summary based on COVID-19 information. We can generate a summary using the GPT-2<\/strong> model from the Hugging Face library.<\/p>\n In this post, we explored how to retrieve information related to Moderna and COVID-19 from Wikipedia using Hugging Face Transformers, as well as the processes of preprocessing and analyzing the data. Hugging Face is a great tool for easily utilizing the latest natural language processing models, enabling more effective utilization of text data. In the future, we can further develop our data analysis skills through various NLP tasks.<\/p>\n Moreover, Hugging Face is continuously updating new models and datasets through collaboration with the community, so ongoing learning and application are encouraged. I hope you will challenge yourselves with diverse NLP tasks and achieve greater achievements.<\/p>\n1.1 Key Features of Hugging Face<\/h3>\n
\n
2. Installation and Environment Setup<\/h2>\n
pip install transformers wikipedia-api<\/code><\/pre>\n3. Retrieving Information from Wikipedia<\/h2>\n
3.1 Example of Retrieving Wikipedia Data<\/h3>\n
import wikipediaapi\n\n # Initialize Wikipedia API\n wiki_wiki = wikipediaapi.Wikipedia('en')\n\n # Retrieve \"Moderna\" page\n page = wiki_wiki.page(\"Moderna\")\n\n # Print page content\n if page.exists():\n print(\"Title: \", page.title)\n print(\"Summary: \", page.summary[0:1000]) # Print first 1000 characters\n else:\n print(\"Page does not exist.\")<\/code><\/pre>\n3.2 Retrieving COVID-19 Related Information<\/h3>\n
# Retrieve \"COVID-19\" page\n covid_page = wiki_wiki.page(\"COVID-19\")\n\n # Print page content\n if covid_page.exists():\n print(\"Title: \", covid_page.title)\n print(\"Summary: \", covid_page.summary[0:1000]) # Print first 1000 characters\n else:\n print(\"Page does not exist.\")<\/code><\/pre>\n4. Text Preprocessing<\/h2>\n
4.1 Preprocessing Steps<\/h3>\n
import re\n\n def preprocess_text(text):\n # Remove special characters\n text = re.sub(r'[^A-Za-z0-9\\s]', '', text)\n # Replace multiple spaces with a single space\n text = re.sub(r'\\s+', ' ', text)\n return text.strip()\n\n # Example of preprocessing\n processed_text_moderna = preprocess_text(page.summary)\n processed_text_covid = preprocess_text(covid_page.summary)\n\n print(\"Processed Moderna Text: \", processed_text_moderna)\n print(\"Processed COVID-19 Text: \", processed_text_covid)<\/code><\/pre>\n5. Analyzing Information with Hugging Face Transformers<\/h2>\n
5.1 Using BERT Model<\/h3>\n
from transformers import BertTokenizer, BertModel\n import torch\n\n # Load BERT model and tokenizer\n model_name = 'bert-base-multilingual-cased'\n tokenizer = BertTokenizer.from_pretrained(model_name)\n model = BertModel.from_pretrained(model_name)\n\n # Tokenize text and convert to tensor\n inputs = tokenizer(processed_text_moderna, return_tensors='pt', padding=True, truncation=True)\n\n # Feed into model and extract features\n with torch.no_grad():\n outputs = model(**inputs)\n \n # Feature vector\n embeddings = outputs.last_hidden_state\n print(\"Embedding Size: \", embeddings.shape)<\/code><\/pre>\n6. Practice Example: Summarizing COVID-19 Related Documents<\/h2>\n
6.1 Summarizing with GPT-2 Model<\/h3>\n
from transformers import GPT2Tokenizer, GPT2LMHeadModel\n\n # Load GPT-2 model and tokenizer\n gpt2_model = GPT2LMHeadModel.from_pretrained('gpt2')\n gpt2_tokenizer = GPT2Tokenizer.from_pretrained('gpt2')\n\n # Input text for summarization\n input_text = \"COVID-19 is caused by SARS-CoV-2...\"\n input_ids = gpt2_tokenizer.encode(input_text, return_tensors='pt')\n\n # Generate summary\n summary_ids = gpt2_model.generate(input_ids, max_length=50, num_beams=5, early_stopping=True)\n summary = gpt2_tokenizer.decode(summary_ids[0], skip_special_tokens=True)\n\n print(\"Generated Summary: \", summary)<\/code><\/pre>\nConclusion<\/h2>\n
References<\/h2>\n