7 NLP Projects for All Levels

One of the best ways to land a job in the field of data science is to build a portfolio with data science projects that effectively show your technical skills. With the boom of ChatGPT, showing the recruiter that you can solve NLP problems has become more important than ever.

In this article, I will show you seven examples of NLP projects for all levels, from the aspiring data scientist to the experienced professional. Let’s get started!

Looking to improve your NLP skills? Start our Natural Language Processing in Python Track today. 

Why Start an NLP Project?

There are a lot of reasons why you should try to solve an NLP task. The first is the market demand. Large Language Models (LLMs), like ChatGPT, captured the attention of all kinds of organizations, meaning they want to invest in these new tools and need people who can demonstrate an understanding of natural language processing.

Furthermore, an NLP project can help you:

• Learn and add a new skill to your CV.
• Build a portfolio of projects that demonstrate your skills and your ability to solve a different range of tasks.
• Show that you keep updated about the new advancements.

NLP Projects for Beginners

These NLP projects are for people starting their data science journey. In these projects, you can master NLP basic concepts, like text processing techniques, bag-of-words, and tf-id.

If you need a refresher on NLP, you can check out our Introduction to Natural Language Processing in Python Course. It can also be helpful to take our Supervised learning with scikit-learn Course to learn machine learning techniques to solve supervised problems.

1. Extract stock sentiment from news headlines

Sentiment analysis is one of the most popular NLP projects. It consists of predicting if a piece of text is positive, negative, or neutral. Understanding the sentiment can bring insights for your business to monitor if there is satisfaction/dissatisfaction with your products.

In the Extract Stock Sentiment from News Headlines project, you will train a sentiment analysis model on the financial news headlines from Finviz. First, you’ll clean the text, and then you’ll apply machine learning techniques to detect if there is a good feeling about the stock or not.

An example from this NLP project

2. Who's Tweeting? Trump or Trudeau?

Another popular project is the data analysis of tweets since Twitter allows to download data using its robust API.

In the Who’s Tweeting? Trump or Trudeau project, you will classify if the tweet is written by Donald Trump or Justin Trudeau. Compared to the previous project, extracting information from tweets can be more challenging because they are short and full of mentions, emojis, and hashtags.

Intermediate NLP Projects

After learning text cleaning, processing, visualization, and application of machine learning models for classification tasks, it’s time to pass to the next level. In the following projects, you will learn three different applications of natural language processing: topic modeling, named entity recognition, and recommendation systems.

3. The Hottest Topics in Machine Learning

NLP techniques aren’t just limited to dealing with labeled datasets; they can also solve unsupervised problems. Topic modeling is one of the main applications for its ability to extract the most representative topics in a collection of documents, like reviews regarding products.

In the Hottest Topics in Machine Learning project, you will discover topics from research papers of NIPS, which is a prestigious machine learning and computational neuroscience conference held every year. The project can be divided into two parts: the pre-processing step and the identification of topics using the Latent Dirichlet Allocation (LDA).

An example from the Hottest Topics in Machine Learning NLP project

4. Resume analysis using Spacy

Named Entity Recognition is a task of Natural Language Processing that consists of identifying and classifying named entities present in a text document into predefined categories, such as person, organization, location, and date.

In the Resume Analysis using Spacy project, you will build a system that helps recruiters to manage effectively the CVs of candidates based on skills that are necessary for the job. The dataset is a collection of resumes taken from livecareer.com. In this project, the spaCy model will be used for recognizing entities in the resume.

5. Book recommendations from Charles Darwin

We are influenced by recommendation systems every day. When you buy a product on Amazon, you can see suggestions for products based on your tastes. The same happens when you watch a film on Netflix, and you have a list of movies based on past choices.

In the Book Recommendations from Charles Darwin project, you will build a book recommendation system based on their content. The data was taken from Project Gutenberg. Charles Darwin’s bibliography will be utilized to identify the books that might capture your interest.

Advanced NLP Projects

The data science projects focus on solving more advanced problems, like language translation and question-answering. You will train models based on transformers to solve each task.

6. English/Italian translator with Hugging Face model

Every year, language translation is becoming better and more accurate. This advancement is thanks to the development of sophisticated language translation techniques.

In the English/Italian Translator with Hugging Face model project, you will build your own translation application with Hugging Face, which is an AI platform that hosts a lot of large language models specialized in different tasks, including language translation. In this project, you pick this model to translate the text from Italian to English. This application is concretized using Streamlit.

7. Question answering with a fine-tuned BERT

Large language models, like ChatGPT, have brought enthusiasm to solving a huge variety of NLP tasks, including question answering. Asking a question and obtaining an answer quickly from a large language model can really speed up the work of people and focus on other challenging tasks.

In the Question Answering with a fine-tuned BERT project, you will fine-tune BERT on the CoQA dataset, which consists of a collection of 127 thousand questions with answers released by Stanford in 2019. The goal is to use the BERT model to answer questions based on the dataset provided.

Conclusion

That’s it! With these projects, you will acquire new skills and enrich your portfolio with NLP projects, which will make you more interesting to the recruiter who is searching for new talents. Based on the level, you can choose the project you feel is more suitable.

If you are interested in getting started with Natural Language Processing, the best way is to take a look at DataCamp’s Natural Language Processing in Python track. You can also check the Natural Language Processing tutorial.