Need a Model for Extracting Relevant Keywords for Given Titles

Hello Hugging Face community,

I have a list of different titles, and I need a model that can take each title and generate at least 30 relevant keywords associated with it. The keywords should be individual words, related to the title, and prioritized based on relevance, with a preference for keywords that are trending or have high search volume. The goal is to extract keywords that best summarize the content or theme of the title.

For example, if the title is:
“A butterfly landing on a dew-covered flower,”
The model should generate keywords like:
butterfly, flower, dew, nature, landing, wings, garden, insect, vibrant, sunlight, spring, outdoors, wildlife, etc.

While I prefer to generate keywords based on search trends or what is currently popular, I am open to models that generate keywords relevant to the title, even if they may not necessarily be trending.

I prefer to work with Python for implementing the solution, and I am looking for a model that can handle this efficiently and output 30 keywords per title.

Could anyone suggest a suitable model for this task, or if possible, provide guidance on how to proceed?

Thank you in advance for your help!

1 Like

If you give detailed instructions to a general-purpose LLM with at least 8B parameters using a prompt, you should be able to get the word list you want.
You can try LLM for yourself using the following.
The problem is that often unnecessary data is included, or the contents are confusing, so you may need to fine-tune the model or organize the output data yourself to make it accurate.
I’m not familiar with fine-tuning…:sweat_smile:

All HF models can basically be used from Python.